Introduction

Fine-grained (pelitic) sediments dominate the soil layering of various Alpine basins and valleys. Many of these basins were formed during several glacial–interglacial periods of the Pleistocene and Holocene and remained as lakes after the last glaciation (van Husen 1979). Lacustrine sediments have filled the basins over thousands of years and are usually characterized by high groundwater tables nowadays. These sediments can lead to an increased risk for building settlements due to their moderate stiffness and strength properties (Petry and Little 2022), leading to high costs for foundation and construction works due to large or differential settlements (Marte and Oberhollenzer 2022). In contrast, it was observed at different construction sites that soft postglacial deposits may be characterized by an unexpectedly high stiffness, leading to small building settlements on shallow foundations. Therefore, geotechnical engineers often assume that most (pelitic) soils have a distinct structure (i.e. defined as the combination of particle arrangement and bonding), which may lead to increased soil stiffness and strength properties.

Soil particle characteristics (e.g. size, shape, mineralogical composition), as well as their arrangement and interparticle bonds, significantly influence soil compressibility, strength and permeability (Leroueil and Vaughan 1990; Mitchell 1976). Here, the term fabric is used to describe the arrangement of particles and pore spaces in pelitic soils. Mitchell (1976) and Burland (1990) introduced the term structure to describe the effects of particle arrangement (fabric) and bonding. Consequently, soils that are characterized by the same fabric can exhibit different properties in terms of stiffness and strength due to different degrees of interparticle bonding. Significant modifications of the original soil structure can, for example, result from the use of inadequate sampling techniques, leading to a reduction in stiffness and strength of the destructed (or non-structured) soil compared to the natural state (Cotecchia and Chandler 2000). While the initial structure is fully lost for remoulded specimens, artificially prepared specimens are named reconstituted (Fearon and Coop 2000). Indeed, natural and reconstituted soils differ significantly with respect to their fabric and particle bonding since compositional factors (e.g. soil particle mineralogy, size and shape), environmental factors (e.g. pore water composition, pressure, temperature and weathering regimes), chemical processes (e.g. leaching, cementation, oxidation and mineral alteration), physical processes (e.g. consolidation, freezing and thawing) and biological processes (e.g. microbial activity, burrowing and production and decay of organic matter) influence the soil`s structure. Consequently, the origin and formation of soil structure may differ between locations (Leroueil and Vaughan 1990), although Tavenas and Leroueil (1980) argue that most natural soils exhibit some structure.

An improved characterization of soil structure(s) is therefore required (i) to capture the mechanical properties of natural soils and assess their changes during construction, (ii) to optimize existing construction measures (e.g. excavation support and foundation systems) and (iii) to introduce sustainable construction concepts, such as long-term carbon binding and sequestration (e.g. shallow carbon storage) in soft soils. While the structure of marine and lacustrine sediments was studied at various locations (Cotecchia et al. 2007, Tsige et al. 1994, Nespereira et al. 2021), only a few investigations have been executed in normally consolidated Alpine deposits. In this article, the degree and origin of soil structure development in pelitic sediments are investigated at three test sites in Austria, namely Lokalbahn Salzburg, Rhesi and water reservoir Raggal, using in situ and laboratory testing based on recovered soil and pore water samples.

Materials and methods

In situ testing

Piezocone penetration tests (CPTu) were used for site characterization, including soil profiling at the three test sites. During test execution, the tip resistance qc, the sleeve friction fs and the dynamic pore water pressure u2 were measured. The dynamic pore water pressure u2 was obtained at the shoulder of the cone. A 15 cm2 cone with a tip opening angle of 60° was pushed into the soil at a constant penetration rate of 2 ± 0.5 cm/s using a hydraulic unit in combination with 1–m-long penetration rods. The testing procedure is standardized following ISSMGE (IRTP 1999) and several national standards (e.g. ISO 22476-1 2012, ASTM D5778-12 2012). In the present work, all CPTu were executed according to ISO 22476-1.

As qc is affected by u2, generated at the shoulder of the cone, Jamiolkowski et al. (1985) suggested using u2 to determine the corrected tip resistance qt based on Eq. (1):

$$q_{t} = q_{c} + \left( {1 - a} \right) \cdot u_{2}$$
(1)

where a is the cone area ratio, usually ranging between 0.8 and 0.9. Since the in situ stress state strongly influences the in situ measurements, Robertson (2009) introduced the updated normalized tip resistance Qtn and the normalized friction ratio Fr for CPTu-based soil classification as summarized by Eqs. (2) and (3):

$$Q_{tn} = \frac{{\left( {q_{t} - \sigma_{v0} } \right)}}{{p_{a} }} \cdot \left( {\frac{{p_{a} }}{{\sigma^{\prime}_{v0} }}} \right)^{n}$$
(2)
$$F_{r} = \frac{{f_{s} }}{{q_{t} - \sigma_{v0} }}100{\text{\% }}$$
(3)

where σv0 and σ'v0 are the in situ total and effective vertical stresses, respectively, and pa represents the atmospheric pressure. The stress exponent n was calculated based on the soil behaviour type index IC (Robertson 2016).

Direct push downhole seismic tests (DPDH) were performed by penetrating a seismic module, equipped with geophones, into the soil. The advanced system by Marchetti et al. (2008), which consists of two geophones installed at a vertical distance of 50 cm, was mounted between the penetration rods and the flat dilatometer blade (DMT) and allowed for measuring the shear wave velocity VS. Therefore, penetration was stopped at regular intervals for VS measurements. The seismic wave, triggered by a hammer blow and received by both geophones, was amplified, digitized and send to the computer for data interpretation (Marchetti et al. 2008). The true interval interpretation method, described in more detail by Marchetti et al. (2008), was used to determine VS. For each testing depth (~ 50 cm intervals), three repeated seismic measurements were performed and averaged for data interpretation. The individual VS measurements at defined testing depths differed by less than 5% (ASTM D7400/D7400M-19 2019).

Various authors (e.g. Goto et al. 1992; Nishi et al. 1989) confirmed that microstructure reinforces the bonds between soil particles, leading to an increased VS and small-strain stiffness shear modulus G0 compared to unstructured soils. Eslaamizaad and Robertson (1996) and Schnaid (2009) suggested linking G0/qt and Qtn to identify the soil microstructure by means of SCPTu. The increase in G0 due to microstructure development is more pronounced than the increase in tip resistance (qt, Qtn) since the latter is significantly affected by destructuration, i.e. the loss of microstructure during penetration (Robertson 2016). This is also confirmed by numerical analyses of cone penetration in fine-grained structured soils (Hauser and Schweiger 2021). Robertson (2009) introduced the modified normalized small-strain rigidity index K*G, determined according to Eq. (4):

$$K_{G}^{*} = \left( {\frac{{G_{0} }}{{q_{n} }}} \right) \cdot \left( {Q_{tn} } \right)^{0.75}$$
(4)

where qn (= qt − σv) defines the net tip resistance. Soils characterized by K*G < 330 are classified as ideal soils with no or little microstructure, whereby young unstructured soils tend to K*G ≈ 100. Soils with K*G > 330 are structured according to Robertson (2016).

Soil sampling

Soil sampling was performed at the test sites Rhesi and Lokalbahn Salzburg based on core drillings, using a single core system without water flushing. PVC tubes with a length of 600 mm (Lokalbahn Salzburg) and 400 mm (Rhesi) were pushed into the soil, retrieved and sealed at both ends with paraffin. Tubes with an outer diameter of 100 mm or 125 mm and a wall thickness of 3.5 mm were used. Offshore sampling at the test site water reservoir Raggal was executed using the CPT-Ranger system (height = 50 cm, diameter = 7 cm) (Geomil equipment 2021). Foam-lined wooden boxes were used to reduce sample disturbance during transport to the laboratory.

Laboratory testing

VS was further determined by means of bender element testing (Arroyo 2001). Here, parallel-type bender elements (length = 12 mm, width = 12 mm, thickness = 1.2 mm) with polyurethane coating were used within a triaxial device (Arroyo et al. 2010). Soil specimens with a diameter of 100 mm and a height ranging from 180 to 200 mm were investigated. A sine wave (amplitude = 12 V) with frequencies ranging between 1 to 15 kHz was considered for arrival time detection. VS is calculated according to Eq. (5) based on the tip-to-tip travel length Ltt (between the emitting and receiving bender elements) and the travel time t of the shear wave.

$$V_{S} = \frac{{L_{tt} }}{t}$$
(5)

No standardized procedures are yet available for defining the arrival time of shear waves (e.g. Arroyo et al. 2010; Lee and Santamarina 2005; Wang et al. 2017). Many authors recommend using the first reversal or the zero after the first reversal as shear wave arrival time (Arroyo et al. 2010; Lee and Santamarina 2005). Here, the zero after the first reversal was used to calculate the arrival time.

VS of the (presumably) structured material, measured in situ by means of SDMT (VS,SDMT), and in the corresponding unstructured material, determined through bender element testing (VS,BE) on reconstituted samples, is compared to quantify the degree of microstructure. Recovered soil specimens are not used for the comparison with reconstituted specimens, since the in situ structure was slightly reduced during sampling. While for unstructured soils VS,BE/VS,SDMT ≈ 1 is expected, smaller ratios are obtained in structured soils (Fonseca and Pineda 2017). Soil specimens considered for bender element testing were oven-dried at 60 °C in the first step. Subsequently, the dried material was mixed with distilled water to reproduce the in situ water content (ranging between 75 and 90% of the liquid limit). The soil–fluid mixture was smeared into cylindrical shells (diameter = 100 mm, height = 220 mm), while ensuring a degree of saturation Sr > 98%, and preloaded vertically (σ′v,applied = σ′v,in situ − 20 kN/m2) without lateral deformation. After preloading, soil specimens (height ranged between 170 to 190 mm, diameter = 100 mm) were installed into a triaxial device and a K0 consolidation procedure was defined to reach the in situ stress state. The in situ stress state was estimated based on the unit weight (determined in the laboratory) and the friction angle (assumed based on experience). Subsequently, VS (propagated vertically through the soil specimen) was measured by means of bender elements. Due to the fine-grained nature of the investigated (pelitic) sediments, any potential effect of the soil’s anisotropy on VS data is negligible, which ensures direct comparison of the field and laboratory data.

The mineralogy of the soil samples was determined through quantitative X-ray diffraction (XRD) analyses (Moore and Reynolds 1997) using a PANalytical X'Pert PRO diffractometer (Co-Kα radiation source) operated at 40 kV and 40 mA and outfitted with a high-speed Scientific X'Celerator detector. The soil powders were prepared by the top-loading technique (Baldermann et al. 2021) and examined in the range of 4–85° 2θ with a step size of 0.008° 2θ and a scan speed of 40 s per step. For data interpretation, the PANalytical X'Pert Highscore Plus software and the ICSD database were applied with an analytical error of < 3 wt%.

The major element composition of the soils was determined by a PANalytical PW2404 wavelength dispersive X-ray fluorescence (XRF) spectrometer using standard glass tablets that were prepared in a fully automatic PANalytical Perl'X 3 bead preparation system. Gravimetric analyses of powdered samples (2 g heated to 1050 °C for 1 h) were used to determine the loss on ignition (LOI) (Baldermann et al. 2020a). The chemical index of alteration (CIA) was calculated based on the XRF results, which presents a widely accepted chemical weathering indicator (Bahlburg and Dobrzinski 2011; Goldberg and Humayun 2010; Nesbitt and Young 1982).

The soil (micro)fabric was studied on representative, sampled soil specimens that were cut perpendicular to the bedding plane using scanning electron microscopy (SEM). Secondary electron images provide information about the particle shape, size and pore system, while energy-dispersive X-ray spectroscopy (EDX) data were collected to chemically fingerprint distinct mineral phases. For further information, reference is made to Collins (1983). Here, Zeiss Sigma 300 VP and Zeiss DSM 982 Gemini microscopes were used for microstructure characterization, which are available at the ZFE and University of Graz, respectively.

The total organic carbon (TOC) and the soil organic matter (SOM) contents were used to describe the influence of biological activity on soil structure formation. Powdered samples were catalytically combusted and the liberated carbon dioxide CO2 analysed by a non-dispersive infrared detector using a Shimadzu TOC-VcPH + ASI-V analyser with an analytical error of ± 5% (Baldermann et al. 2020a). The SOM content was calculated based on TOC values by assuming a conversion factor of 2.1 (i.e. SOM = TOC ⋅ 2.1).

The pore water of the soil specimens was characterized with respect to the electric conductivity (EC), pH value and elemental concentration. EC was determined based on a WTW LF 330 m multimeter connected to a TetraCon 325 probe. A WTW pH 330 m multimeter in combination with a WTW SenTix41 probe was used to analyse pH. The pH electrode was calibrated against NIST buffer standard solutions at pH = 7 and pH = 10 at 23 °C (Steindl et al. 2019) with an analytical precision of ± 0.03 pH units. Major, minor and trace element analyses were performed on a Perkin Elmer Optima 8300 DV inductively coupled plasma optical emission spectrometer (ICP-OES) on acidified sample aliquots as well as via ion chromatography (Dionex ICS-3000) and automated titration (Schott TA20plus titrator) at an analytical precision of < 5% (determined via NIST 1640a, in-house and SPS-SW2 Batch 130 standards). The saturation index SI, defined according to Eq. (6), was used to determine whether a mineral is saturated (SI = 0), undersaturated (SI < 0), or supersaturated (SI > 0) with respect to the pore water:

$${\text{SI}} = \log \frac{{{\text{IAP}}}}{{K_{sp} }}$$
(6)

where IAP and Ksp represent the ion activation product and the mineral-specific solubility constant (e.g. quartz, calcite and mica), respectively.

The age of the investigated soils was determined by radiocarbon dating (C-14) of organic components present in the soils (e.g. wood sticks or plant remnants), with recognition of the published radiocarbon decay constant (Garrett 1963). All measurements were executed at the Poznań Radiocarbon Laboratory using accelerator mass spectrometry (AMS).

Results and discussion

Soil classification

Lokalbahn Salzburg

The test site Lokalbahn Salzburg is located in Salzburg city (47° 48′ 58.8492'' N/13° 2′ 42.2592″ E) and in situ testing (CPTu, SDMT) as well as soil sampling was executed down to a depth equal to 25 m. The subsurface sediments can be divided into three main lithologies based on CPTu measurements, denoted CPTu 8/20 and CPTu 19/20 in the following, i.e. L1—sandy gravel, L2—sand-silt alternations and L3—silty clay. The groundwater table was situated 1.5 m below ground level during test execution. L1 is characterized by a thickness of 4 m and the measured qc (5 to 20 MPa) and fs (20 to 160 kPa) values show a large scatter, indicating a heterogeneous composition (Fig. 1a, b). Small Rf (< 1%) and u2 values (Fig. 1c) imply drained material behaviour during in situ testing. These sediments are classified as (6) sands (clean sand to silty sand) according to Robertson (2009) (blue symbols in Fig. 2a) and silty gravel with sand (GM) based on ASTM D2487-17e1 2017 (USCS). Sand-silt alternations (L2) are situated between 4 to 9 m below ground level and exhibit reduced qc and fs values compared to L1, ranging from 0.8 to 4.0 MPa and 10 to 80 kPa, respectively. The erratic patterns of u2 is consistent with varying soil behaviour types (red symbols in Fig. 2a: (3) clays to (5) sand mixtures). Although the PSD varies within L2 (fines-fraction varies between 19.2% and 97.4%), silt and sand remain the dominant fractions according to ISO 14688-1 (2017) (Fig. 3a: cl’ sa Si to si Sa). The homogeneous section L3 is characterized by reduced qc (< 2 MPa) and fs (< 15 kPa) values, high excess pore water pressures at position u2 and VS values ranging from 200 to 300 m/s (Fig. 1a–d). These sediments are classified as (3) clays according to Robertson (2009) (green symbols in Fig. 2a), sandy clayey silts (sa’ cl Si) according to ISO 14688-1 and lean clays (CL) according to USCS (ASTM D2487-17e1). The Atterberg limits show a small scatter (PL = 19.5 to 23.6%, LL = 39.7 to 42.2%; Fig. 3b), and all investigated specimens are classified as medium plasticity clays based on the Casagrande chart. As shown in Fig. 3b–d, wnat, ρd, ρsat and ρs vary from 25.3 to 37.7%, 1.41 to 1.65 g/cm3, 1.94 to 2.09 g/cm3 and 2.77 to 2.78 g/cm3, respectively. The SOM contents vary between 0.3 and 0.9%, but show no significant trend over depth. The soil age was investigated at two depths (− 12 m and − 20 m). It is evident that the sediments were deposited in the Pleistocene, at 32,400 ± 500 and 36,100 ± 700 years before present (BP).

Fig. 1
figure 1

Compilation of CPTu and DPDH measurements (qt, fs, u2, VS) at test sites Lokalbahn Salzburg (top row), Rhesi (middle row) and water reservoir Raggal (bottom row)

Full size image
Fig. 2
figure 2

Soil classification according to Robertson (2016) for test sites a Lokalbahn Salzburg, b Rhesi and c water reservoir Raggal

Full size image
Fig. 3
figure 3

Compilation of results from geotechnical laboratory testing (particle size distribution PSD, liquid limit LL, plastic limit PL, dry density ρd, saturated density ρsat, particle density ρs) at test sites Lokalbahn Salzburg (top row), Rhesi (middle row) and water reservoir Raggal (bottom row)

Full size image

Rhesi

The test site Rhesi is situated in the western part of Austria about 2 km south of Lake Constance (47° 28′ 38.4924″ N/9° 40′ 5.4912″ E). All investigations were executed within a rectangular area of approximately 75 × 25 m down to a depth of 25 m below ground level. The subsurface at the test site Rhesi can be divided into three lithologies: L1—silty sand, L2—sand-silt alternations and L3—silty clay (see Fig. 1e–h). L1 represents a 2 m thick top layer that is characterized by a large heterogeneity, leading to a strong variation of CPTu measurements (Fig. 1) and SBTn classifications (blue symbols in Fig. 2b). The groundwater table was situated 0.8 m below ground level during test execution. While qc and fs are smaller than 1 MPa and 10 kPa in fine-grained sections of lithology L2, they can increase up to 5 MPa and 30 kPa in coarser sections. The reverse trend is observed for u2 measurements. Since Rf varies between 1 and 3%, the data yield a wide scatter within the SBTn chart of Robertson (2009), ranging from (3) clay to (6) sand (red symbols in Fig. 2b) and generally indicating normally consolidated conditions (Robertson 2009). These sediments present a fines-content ranging from 33 to 92% and are classified as lean clays (CL) to silty, clayey sands (SC-SM) based on USCS (ASTM D2487-17e1). All data are situated above the A-line within the Casagrande chart, even though the particle size distribution is characterized by a pronounced scatter (ranging from sa' cl Si to cl' si Sa according to ISO 14688-1, Fig. 3e). The sediments of L3 (Fig. 1: qc < 1 MPa, fs ≈ 10 to 15 kPa and Rf = 1 to 1.5%) are characterized as (3) clays according to Robertson (2009) (green symbols in Fig. 2b), present a fines-content larger than 91% (lean clays CL to fat clays CH according to USCS; sa’ cl Si to sa’ Cl/Si according to ISO 14688-1) and show a slight increase of LL and PL over depth (triangles in Fig. 3f). The values for wnat, ρd, ρsat and ρs vary from 34 to 38% (Fig. 3f), 1.28 to 1.38 g/cm3 (Fig. 3g), 1.8 to 1.86 g/cm3 (Fig. 3g) and 2.64 to 2.69 g/cm3 (Fig. 3h), respectively, within L3. The SOM contents vary over depth and present the highest values within L3 (0.4 to 1.1%). Age dating of sediments collected at − 3 m and − 17 m depth indicates that sediment deposition occurred in the Holocene, between 3,800 and 4400 ± 40 years BP.

Water reservoir Raggal

The test site water reservoir Raggal, located in the western part of Austria (47° 13′ 13.2024″ N/9° 50′ 38.0724″ E), is operated by the energy supplier illwerke vkw and is part of the hydropower plant Oberstufe Lutz (Oberhollenzer et al. 2022b). Due to natural sedimentation, about 50,000 m3 of material is deposited within the reservoir every year. In situ testing and soil sampling were executed from a floating pontoon using a modular stand-alone system (Fox-150) by Geomil equipment (2022) as pushing device. The water depth amounted to approximately 7 m during test execution. The sediments were investigated down to approximately 15 m below seabed. The sedimentation history within the water reservoir Raggal is reflected in the form of alternating fine- and coarse-grained layers. The sand-silt alternations are characterized by an erratic distribution of CPTu patterns over depth (L1 in Fig. 1i–k). Values for qc and fs vary from 0.7 to 3.5 MPa and 5 to 50 kPa, respectively. VS varies between 50 and 230 m/s over depth and presents increased values within sand-dominated sections (Fig. 1l). These young deposits, characterized by an age of < 50 years, are mainly defined as (4) silt and (5) sand mixtures based on Robertson (2009) (red symbols in Fig. 2c). The PSDs of soil specimens reveal that an increased sand content (e.g. at − 10.5 m) correlates with higher qt and lower u2 values (Figs. 1i, 3i). Four soil samples, classified as clayey, sandy silts (cl’ sa’ Si) according to ISO 14688-1, were recovered within fine-grained sections. These specimens present a fines-content of > 90% and are classified as organic silts (OL, OH) based on the USCS. Due to their high SOM contents, ranging between 0.63 and 2.1%, all data are situated below the A-line within the Casagrande chart. Small ρd from 1.19 to 1.23 g/cm3 and ρsat from 1.74 to 1.77 g/cm3 and a high wnat of > 40% were determined within fine-grained layers (Fig. 3). The ρs slightly decreases from 2.71 to 2.61 g/cm3 over depth (Fig. 3l).

Seismic in situ testing

SBTn charts according to Robertson (2016) are shown for all three test sites in Fig. 4. The results obtained for test site Lokalbahn Salzburg are situated at the transition between unstructured and structured soils (Fig. 4a). While most samples from L1 (sand gravel) and L2 (sand-silt alternations) fall within K*G > 330, the samples from L3 (pelitic sediments) are characterized by slightly smaller K*G values (K*G ≈ 330, on average). However, individual K*G values of L3 are significantly larger than 330 (Fig. 4a). These results are in good agreement with previous investigations executed at construction sites in Salzburg city (Oberhollenzer et al. 2019).

Fig. 4
figure 4

Detection of microstructure at test sites a Lokalbahn Salzburg, b Rhesi and c water reservoir Raggal according to Robertson (2016)

Full size image

SBTn results are situated at the transition between ideal (unstructured) and structured soils for test site Rhesi (Fig. 4b). In contrast to the previous test site, the top layer (L1) and sand-silt alternations (L2) are characterized by K*G < 330. Since L3 (clayey silt) shows a bit higher K*G values, soil microstructure slightly increases over depth according to Robertson (2016). However, K*G remains smaller than 330 within L3. Fine-grained deposits investigated at test site water reservoir Raggal are again situated at the transition between structured and unstructured soils (Fig. 4c). While sand-dominated layers are characterized by K*G > 330 (see 4 data points situated on the right side of the K*G = 330 line in Fig. 4c), K*G values smaller than 330 are observed in silt-dominated sections.

Since the microstructure of pelitic sediments is investigated in detail, the in situ data were filtered based on the SBTn to enable a comparison of fine-grained layers investigated at the different test sites. Sediments characterized by SBTn = 3 (Lokalbahn Salzburg, Rhesi) and SBTn = 4 (water reservoir Raggal) are used for further discussion. The filtered qt, fs, Qtn and G0 values are plotted against σ’v for a direct comparison in Fig. 5. It is shown that qt and fs are in good agreement between test sites Lokalbahn Salzburg and Rhesi, especially in greater depths. While the increase in qt with depth correlates with higher effective stresses, fs seems to be less sensitive to changes in stress level, and Qtn remains approximately constant (due to the normalization). However, G0 is higher at test site Lokalbahn Salzburg, which is related to the higher density (Fig. 3c). While CPTu and SDMT were executed to a final depth of 25 m at both locations, CPTu measurements for test site water reservoir Raggal are only available for σ’v smaller than 75 kPa, making a direct comparison with the other sites difficult. The slightly smaller clay content and higher sand content lead to an increase of qt and Qtn compared to test sites Lokalbahn Salzburg and Rhesi (Fig. 5a, c). G0 results increase over depth and are in good agreement with the two other locations (Fig. 5d).

Fig. 5
figure 5

Comparison of the corrected tip resistance qt, sleeve friction fs, updated normalized tip resistance Qtn and small-strain stiffness shear modulus G0 for test sites Lokalbahn Salzburg, Rhesi and water reservoir Raggal

Full size image

K*G results, calculated based on the filtered Qtn and G0 values, are shown in Fig. 6 using box plots. While the oldest sediments, investigated at the test site Lokalbahn Salzburg (age: 32,400—36,100 years BP), lead to the highest K*G, the smallest K*G is obtained within the water reservoir Raggal (age: < 50 years). However, the differences in K*G are small between the three locations, and all results are situated close to the proposed transition between ideal and structured soils (K*G = 330).

Fig. 6
figure 6

Comparison of the modified normalized small-strain stiffness rigidity index K*G for test sites Lokalbahn Salzburg, Rhesi and water reservoir Raggal by means of box plots

Full size image

Comparison of in situ and reconstituted shear wave velocity

The VS data obtained from bender element testing and SDMT are shown in Fig. 7 (top row) for the three test sites. Additionally, the VS,BE/VS,SDMT ratios are presented in the bottom row of Fig. 7. It can be observed for the test sites Lokalbahn Salzburg and Rhesi that the bender element results are smaller compared to the in situ measurements, which leads to VS,BE / VS,SDMT ratios smaller than 1, indicating the presence of a microstructure. Although the ratios are characterized by some scatter, a value of approximately 0.8 is reached at both test sites, which does not change significantly over depth (VS,BE/VS,SDMT ≈ 0.7–0.9; Fig. 7d and e). At the test site water reservoir Raggal, VS,BE/VS,SDMT ratios of approximately 1 are observed, indicating little if any microstructure within the young deposits (Fig. 8f). Consequently, the presence of soil structure is indicated at 2 out of 3 test sites based on the obtained VS,BE/VS,SDMT ratios. In the following, possible explanations for microstructure formation are discussed based on mineralogical, hydrochemical and microstructural laboratory testing.

Fig. 7
figure 7

Comparison of in situ and reconstituted shear wave velocities (VS,SDMT, VS,BE) for test sites Lokalbahn Salzburg (left column), Rhesi (middle column) and water reservoir Raggal (right column). While white triangles are determined based on averaged SDMT readings, the scatter in VS,BE/VS,SDMT is indicated by dotted horizontal lines

Full size image
Fig. 8
figure 8

Compilation of X-ray diffraction (XRD) data for test sites a Lokalbahn Salzburg, b Rhesi and c water reservoir Raggal

Full size image

Mineralogical and chemical characterization of soils

The mineralogy at test sites Lokalbahn Salzburg, Rhesi and water reservoir Raggal is mainly composed of quartz, clay minerals, carbonates (calcite and dolomite) and feldspar (K-feldspar and albite), which differ in quantity between the locations and along vertical profiles. At test site Lokalbahn Salzburg (Fig. 8a), the amount of quartz decreases from 40.4% to 16.2% over depth, while clay minerals present the inverse trend. The latter group is mainly composed of muscovite (19.9–42.6%) and, to a smaller extent, chlorite (3.7–6.6%) and kaolinite (1.2–3.0%). Both, calcite and albite amount to approximately 13% and show no significant changes over depth. The mineralogical composition at test site Rhesi (Fig. 8b) is dominated by quartz (32.8–39.6%), calcite (3.8–45.2%), dolomite (1.2–9.7%) and muscovite (3.3–27.8%). The reduced carbonate content and the high amount of clay minerals within L1 are strong indicators of a different depositional history compared to L2 and L3. This observation agrees with recordings proving that the river Dornbirner Ach (draining limestone and marl lithologies) dominated the early sedimentation at this test site. Later, the Alpine Rhine was artificially diverted to its present river bed. L2 and L3 consist of quartz (32.8–39.6%), muscovite (3.3–17.4%), calcite (30.1–45.2%) and dolomite (4.6–9.7%). Accordingly, test site Rhesi is characterized by higher calcite and lower muscovite contents within L2 and L3 compared to Lokalbahn Salzburg. The mineralogy at test site water reservoir Raggal presents a strong scatter over depth, even though quartz (9.8–26.8%), muscovite (4.3–29.5%), chlorite (0.9–4.6%), calcite (10.8–24.0%), dolomite (19.1–71.7%), albite (0.9–4.0%) and orthoclase (1.0–1.7%) dominate (Fig. 8c).

The CIA values indicate a high to very high degree of weathering, ranging between 76 and 89 at test sites Lokalbahn Salzburg and Rhesi, which is typical for postglacial deposits. The younger sediments at test site water reservoir Raggal show lower CIA values ranging between 67 and 74, suggesting that physical particle breakdown is more efficient than chemical weathering (Nesbitt and Young 1982). It was observed at each test site that the CIA values slightly increase over depth since deeper situated layers are exposed to interactions with groundwater and, thus, chemical weathering for a longer period of time.

Pore water chemistry

The pH values range between 7.4 and 8.2 at all locations, indicating a slightly alkaline environment. The EC values range between 462 and 1430 µS/cm, yielding increased values within the pelitic sections. Therefore, L3 at test sites Lokalbahn Salzburg and Rhesi shows the highest EC values along their respective profiles, but the fine-grained layers at test site water reservoir Raggal exhibited the highest EC values along all test sites (EC = 1430 µS/cm).

The concentrations of Ca2+, Mg2+, K+, Na+, Cl−, NH4+ and SO42− ions within the pore water remain below 160 mg/l within the postglacial deposits. Higher amounts of HCO3− ions can be related to the dissolution of carbonate minerals and natural gas occurrences at test site Lokalbahn Salzburg (163 to 348 mg/l) and Rhesi (237 to 381 mg/l), respectively. The concentrations of Ca2+ (167–364 mg/l), SO42− (142–232 mg/l) and HCO3− (438 to 1086 mg/l) ions vary strongly at test site water reservoir Raggal. The SI values of quartz, aragonite, calcite and albite were calculated from pore water concentration data and are shown in Fig. 9. Near surface layers at test site Lokalbahn Salzburg are characterized by SI ≈ 0 for all minerals of interest (Fig. 9a). In contrast, undersaturation (SI < 0) is indicated for aragonite, calcite and albite within L3. Test sites Rhesi and water reservoir Raggal indicate an oversaturation (SI > 0) with respect to calcite and aragonite in all pore waters (Fig. 9b), suggesting an authigenic formation of these carbonate minerals. While calcite cement formation can be reasonably considered to have a stabilizing effect on soil structure, silicate weathering (indicated by a negative SIAlbite) leads to the reverse trend, as recently documented by artificial soil cementation experiments (Oberhollenzer et al. 2022a). Since silicate weathering needs more time compared to calcite formation, a stabilizing effect and related microstructure formation is expected at test sites Rhesi and water reservoir Raggal.

Fig. 9
figure 9

Saturation indices (SI) calculated for several mineral phases at test sites a Lokalbahn Salzburg, b Rhesi and c water reservoir Raggal

Full size image

Microstructural characterization

The soil fabric of undisturbed pelitic specimens recovered at − 25.3 m, − 22.2 m and − 15.8 m at test sites Lokalbahn Salzburg, Rhesi and water reservoir Raggal is shown in Fig. 10. The presence of lamination is indicated by red shadings for each test site (Fig. 10a, b and c). Since individual soil specimens were rotated for analysis, the red dashed lines are disposed to the horizontal. It is evident that particle sizes differ between the three test sites. While soil specimens recovered at test sites Lokalbahn Salzburg and Rhesi are in good agreement with respect to their PDS (sa’ cl Si), sediments from test site water reservoir Raggal indicate a higher sand content (cl’ sa’ Si). The pore systems (indicated by blue, solid ellipses) are aligned along the lamination at test sites Rhesi (Fig. 10b) and water reservoir Raggal (Fig. 10c). In contrast, no such ‘systems’, i.e. large voids, are observable at test site Lokalbahn Salzburg (Fig. 10a). Since test sites Lokalbahn Salzburg and Rhesi are characterized by similar particle sizes, differences in fabric can be related to changes in ρd and e. Consequently, the open fabric at test site Rhesi (ρsat = 1.81 g/cm3, e = 1.1) is characterized by a lower ρd and higher e than test site Lokalbahn Salzburg (ρsat = 1.95 g/cm3, e = 0.9). The change in ρd is caused by an open fabric within under-consolidated sediments at test site water reservoir Raggal (ρsat = 1.74 g/cm3, e = 1.2). As the oldest and youngest sediments were investigated in Salzburg and Raggal, respectively, it can be seen that soil density (ρd, ρsat) increases, while wnat and e decrease with sediment age (Fig. 3).

Fig. 10
figure 10

SEM images showing fabric and microstructural features of sampled soils from test sites Lokalbahn Salzburg (left column), Rhesi (middle column) and water reservoir Raggal (right column)

Full size image

Calcite (CaCO3), characterized by rhomb-shaped crystals at test site Rhesi (green, dashed polylines in Fig. 10), is likely of authigenic origin, corroborating the pore water chemical data. The calcite cements grow within the pore system and present a pore-infill or interconnect other detrital particles, such as quartz, clay minerals, carbonate clasts and feldspar grains, which both stabilizes the sediment’s fabric. In contrast, no significant amount of authigenic calcite was recognized at test site Lokalbahn Salzburg (Fig. 10d). The sediments investigated at test site water reservoir Raggal present an open fabric, which is characterized by pore systems that are only filled by calcite to a small extent. As shown in Fig. 10f, authigenic calcites are mainly formed on top of existing carbonate grains, indicated as epitaxial growth or overgrowth (Baldermann et al. 2020b).

SEM results of reconstituted soil specimens are shown for the test sites Lokalbahn Salzburg and Rhesi in Fig. 11. The material used for reconstituting the specimens was sampled at − 19.2 m and − 22.2 m, respectively (see Fig. 11). Both specimens do not show lamination and are characterized by a smaller pore space compared to the natural specimens (see Fig. 10). Clay minerals are aligned mostly in parallel (see green indication in Fig. 11) and do not form a card house structure as it is often observed in natural clays. No evidence for the formation of stabilizing (i.e. structure-forming), authigenic calcite cements was found in the reconstituted samples at both sites, all indicating that the fabric differed compared to the sampled specimens.

Fig. 11
figure 11

SEM images showing fabric and microstructural features of reconstituted soils from test sites Lokalbahn Salzburg (left column) and Rhesi (middle column). Note the absence of lamination, the finer pore system and the preferred orientation of clay mineral particles in the reconstituted soils compared to the sampled soil specimens

Full size image

Discussion

Sediment ageing, which is known to significantly influence the mechanical behaviour of soils (Schmertmann 1991), includes (i) structure-forming processes, such as mineral authigenesis, and (ii) structure-destabilizing processes, such as chemical and physical weathering. These two processes are difficult to separate as they may occur simultaneously over time (Burland 1990; Leroueil and Vaughan 1990; Mitchell 1976). The formation of authigenic calcite cements was recognized at two locations (Rhesi, water reservoir Raggal) based on a SEM study and was confirmed by the soil’s pore water chemistry. Micron-sized calcite particles clog formerly open pore spaces and interconnect detrital particles. Accordingly, this cementation process leads to an increased VS (DeJong et al. 2022; Schnaid 2009), which was confirmed by comparing the VS values determined in situ and by means of bender element testing on reconstituted soil specimens. The considered postglacial, pelitic sediments at the test site Rhesi are characterized by VS,BE/VS,SDMT ratios < 1, suggesting the presence of soil structure, which is in good agreement with published seismic CPTu results (Oberhollenzer et al. 2019). The influence of soil structure on results of oedometer and triaxial tests was not investigated in this study, as the specimens used for mechanical testing were slightly disturbed. While an interlocking effect due to cementation was observed at test site Rhesi, an overgrowth of CaCO3 cements on existing detrital calcite surfaces was observed at test site water reservoir Raggal. It is assumed that the latter has only a minor stabilizing effect on soil structure in these extremely young, pelitic soils, since a VS,BE/VS,SDMT ratio of approximately 1 was determined. Since VS,BE/VS,SDMT < 1 are characterized by K*G larger than 280, it is suggested to slightly adjust the transition proposed by Robertson (2016) (i.e. unstructured soils: VS,BE/VS,SDMT ≈ 1, K*G < 280). The results indicate that VS,BE/VS,SDMT ratios are more sensitive to changes in microstructure compared to K*G and are therefore preferred. Silicate mineral weathering, which was observed at all three test sites, leads to the opposing effect, i.e. microstructure destabilization. The latter reactions take longer time spans compared to the authigenic formation of CaCO3 cements and, thus, the structure-forming process is assumed to be dominant for the considered postglacial sediments. Different microstructure-forming mechanisms must be considered as driving force at test site Lokalbahn Salzburg, since no authigenic formation of CaCO3 is indicated based on SEM imaging and pore water chemistry data. Further investigations are required to reveal to what extent additional structure-forming processes and the significantly greater burial age may play a role for the formation of these sediments.

Conclusion

This study investigates the composition and microstructure of lacustrine, fine-grained sediments at three test sites within the Alpine region. While postglacial deposits are studied at the test sites Lokalbahn Salzburg and Rhesi, significantly younger soils (< 50 years) were investigated within the water reservoir Raggal. In situ testing by means of CPTu and SDMT in combination with soil, as well as pore water sampling to enable geotechnical, mineralogical, hydrochemical and microstructural laboratory testing, were executed at each location. The most important findings can be summarized as follows:

  • The modified normalized small-strain rigidity index K*G was used to distinguish structured and unstructured soils. K*G was found to slightly increase with sediment age and degree of calcite cementation, however, not being very sensitive to changes in microstructure.

  • The degree of soil structure was estimated by VS,BE/VS,SDMT: VS,BE/VS,SDMT ratios < 1 indicate the presence of soil structure in postglacial deposits, while younger, unstructured deposits yield VS,BE/VS,SDMT of approximately 1.

  • If the degree of structure is not evaluated based on VS,BE/VS,SDMT, it is recommended to define the transition between unstructured and structured soils at K*G = 280 (at least for fine-grained sediments within the Alpine region).

  • The test sites Rhesi and water reservoir Raggal are characterized by the authigenic formation of rhomb-shaped calcite cements, leading to an interparticle bonding, especially at test site Rhesi. The precipitation of calcite cement interconnects detrital particles and reduces the pore space, thus resulting in higher stiffness properties. No authigenic formation of calcite is observed at test site Lokalbahn Salzburg.

This study highlights the limited applicability of existing soil behaviour type charts for detecting soil structure in continental Alpine basins and emphasizes the great potential of seismic measurements to overcome these limitations. A comprehensive characterization of soil fabric and structure was carried out for Alpine lacustrine sediments, highlighting the importance of cement precipitation as a structure-forming process.

Future investigations should intend to validate and improve existing soil behaviour type charts based on high-quality field data obtained at different locations. Chamber tests should further be used to study the influence of artificial cementation on results of soil behaviour type charts under laboratory conditions. High-quality specimens should be sampled in Alpine basins to characterize the influence of soil structure on mechanical properties based on laboratory tests. We recommend that soil specimens should be characterized further by means of (transmission) electron microscopy performed at high-spatial resolution and novel, mathematical (statistical) algorithms to detect distinct changes in soil mineralogy and pore space distribution (e.g. application of artificial intelligence) with depth and age. Stable isotopes of carbon and oxygen could be investigated to assess the links between CO2 uptake from the atmosphere and the subsequent increase in SOM content and CaCO3 precipitation dynamics. Such multi-analytical approaches will contribute to develop a better understanding of the influence of soil fabric and soil structure on its stiffness and strength properties.