Rheological properties of halloysite soil slurry: a case study of weathered tephra involved in a shallow landslide triggered by the 2018 Eastern Iburi earthquake in Hokkaido, Japan – Earth, Planets and Space

May 11, 2022

Difference in yielding behaviors of slurries formed from dried and moist soil

The present experiments revealed that the yielding and fluidization behavior of slurries of weathered tephra depend greatly on several factors. As mentioned above, the yield stress is greatly affected by the slurry’s water content; despite this, comparison with the liquidity index shows that the yield stresses of both types of slurry are generally lower than those expected from empirical relations (Fig. 5b). One reason for this is that the yield stress for constructing the empirical curves was determined by a dynamic method, i.e., interpolation of a stress–strain rate curve to obtain a shear stress in the limit of zero shear rate, which is commonly higher than the yield stress determined by the static and direct method adopted here.

Creep tests by Carrière et al. (2018) on several clay soils at different IL states determined the critical shear stress at which the slurry viscosity bifurcates with increase of strain. Despite the clay slurries being prepared with dried samples, the stress values agreed well with the empirical relationship described above. Therefore, the low stress values obtained in this study may reflect the characteristics of the present sample, not the experimental method or sample preparation procedure.

An important finding of the present experiment is that the rheological properties of the slurry differ significantly between two different preparation methods (i.e., using dried or moist soil). Several studies have shown that the material properties of soils are altered by heating and drying (Casagrande 1932; Terzaghi et al. 1996; Huvaj and Uyeturk 2018). In general, after oven drying, the Atterberg limits such as liquid limit and plastic limit decrease by several tens of percent. Huvaj and Uyeturk (2018) studied the effect of drying on the Atterberg limit of pyroclastic soils containing halloysite, finding that the limits of samples dried at 60 °C or 110 °C were 1% to 30% lower than those of moist samples; they considered this likely owing to aggregation of the soil fabric or irreversible changes in the properties of halloysite after drying. Figure 5a shows a difference of approximately 0.03 to 0.04 in the scale of φ for similar yield stress values between the two slurries, corresponding to a difference of ~ 40% of the water content ratio. This is equivalent to a ~ 20% decrease in water content (as a proportion of the total moisture content), which is roughly consistent with the findings of Huvaj and Uyeturk (2018). Such differences led the estimated yield stress of the two slurries at the natural moisture state to vary by several orders of magnitude, from 2.2 kPa to several pascals. Kameda and Okamoto (2021) assumed Bingham flow in a numerical model of a landslide triggered by the Eastern Iburi earthquake. The modeling constrained the yield stress of the collapsed soils in a saturated state to be ~ 1.5 kPa, which is of the same order as that of the slurry containing moist soil (i.e., 2.2 kPa), suggesting that such slurry is more likely to represent the soil state when the landslide occurred. In fact, in situ monitoring of the tephra after the earthquake indicated sustained high saturation (> 90%) even during a period of no precipitation (Wang et al. 2021). Similar tephra containing halloysite is likely to be widely distributed in volcanic regions worldwide. If it is sufficiently dried to the extent of the dried soil used in this study and then saturated by rainfall, its yield stress may be reduced significantly, as demonstrated in Fig. 4. It is also expected that the soil strength in such a state will decrease with time rather than immediately after hydration; therefore, these cases will show a time difference between precipitation and the onset of coseismic ground motion due to an earthquake, which will be an important factor in determining slope stability during the earthquake.

Dependence of slurry properties on pH

Our results demonstrate a dependence of yield stress on slurry pH at a given water content. Although the two types of slurry showed differing sensitivity to pH, the yield stress generally decreased under alkaline conditions and increased under acidic conditions (Fig. 6a), as predicted previously from measurements of zeta potential for samples with a finer size fraction (Kameda 2021). Comparison of the two types of slurry showed slurry made of dried soil had a more moderate dependence on pH than that made of moist soil. Zeta potential measurements in the present study used samples of the same size fraction as those used for the rheological tests; the potential showed a near-steady increase from a large negative value (− 40 mV) to near-zero with decreasing pH (Fig. 6b). A large negative zeta potential can enhance colloidal stability, and thus reduce the shear strength of slurry (Plaza et al. 2018). More quantitatively, the following model of yield stress–DLVO (Derjaguin–Landau–Verwey–Overbeeck) force and interaction energy have often been applied to clay slurries based on a constant surface potential for interactions between spherical particles (Scales et al. 1998):

$${tau }_{y}approx frac{{phi }^{2}}{a}left(frac{A}{12{{D}_{0}}^{2}}-2pi {zeta }^{2}frac{{kappa mathrm{e}}^{-kappa {D}_{0}}}{1+{mathrm{e}}^{-kappa {D}_{0}}}right),$$

(6)

where a is the particle radius, A is the Hamaker constant, D0 is the surface separation distance between interacting particles, and (kappa) is the reciprocal of the Debye length (double-layer thickness). This model predicts a linear correlation between the yield stress and the square of zeta potential, and some clay slurries have been found to follow this correlation over a wide pH range (e.g., Au et al. 2014; Au and Leong 2016). As mentioned above, data for the slurry of dried soil show an almost linear correlation between yield stress and the square of zeta potential (Fig. 6c), indicating that the above yield stress–DLVO force model is likely applicable to the pH range tested here. In contrast, the results for the slurry of moist soil did not easily fit a single straight line. This may indicate either a change in the configuration of individual clay particles at different pH conditions (Shankar et al. 2010) or the addition of another attractive force, especially under acidic conditions, where the yield stress increased dramatically (Fig. 6a).

Although the microstructural mechanism for such behavior remains uncertain, the observed pH dependence of yield stress confirmed the prediction of Kameda (2021). Based on a simple mixing experiment of rainwater and tephra sample, Kameda (2021) raised the possibility of a change in pH of pore fluid in the tephra layer during rainfall, with pH likely increasing as the solid/liquid ratio decreases. If so, the chemical environment may have changed due to the rainfall before the earthquake in the present case, reducing the yield stress and thus the stability of the slope, which may have been a factor contributing to slope failure during the earthquake.

Fluidization of slurry under oscillatory strain

At the time of the Eastern Iburi earthquake, the slope was subjected to strong shaking with a maximum seismic intensity of 7 (JMA intensity scale) near the epicenter, which is thought to have directly triggered the slope collapse. To examine the effect of such periodic strain acting on the slope’s weathered tephra, dynamic viscoelasticity tests were conducted at frequencies of 0.5 to 10 Hz (Figs. 7 and 8). The results indicate that both types of initially solid-like slurry can liquify within the tested frequency range. However, the slurry of moist soil had a higher resistance to oscillatory strain than the other slurry, as expected from the vane test, and its fluidizing point occurred at a higher moisture content than in the case of the slurry of dried soil (Figs. 7d and 8d). In all cases, the strain amplitude of the solid–liquid transition (i.e., crossover strain) increased as the water content decreased or the frequency decreased. The slurry of moist soil, which is probably closer to the state of the soil at the time of the earthquake, had its crossover strain greatly increase when the applied frequency was less than 2 Hz, which may have been caused by the increase of storage modulus G′ above a shear strain of ~ 10% (Fig. 8a, b). In this study, experiments at even lower ratios of water content, such as the natural moisture state, could not be performed, because the samples cracked during shearing. However, extrapolation of the experimental data indicates that the phase transition is expected under conditions of natural water content at small strain conditions when the frequency exceeds 5 Hz (Fig. 8d).

Previous geotechnical experiments using the same tephra materials indicated that halloysite-bearing tephra is more susceptible to liquefaction than other horizons under cyclic loading at a frequency of 1.5 Hz (Li et al. 2020). On the other hand, ground motion with strain occurring in several pulses at higher frequency (~ 3 Hz or more) during the earthquake was recorded at several seismological stations near the epicenter (Wang et al. 2019). Recent continuous seismic observation after the Eastern Iburi earthquake revealed that the characteristic frequency of the slope oscillation associated with seismic activity ranged between 5 and 7 Hz (Wang et al. 2020). The experimental results indicate that such oscillatory strain could easily fluidize the tephra layer, while at lower frequencies it may exhibit some resistance to fluidization. Although this study tested sieved soil whose original texture was completely broken, the obtained rheological properties may be a key to understanding the factors contributing to the stability of the slope during the long period after tephra deposition and the mechanism of eventual collapse due to seismic shaking.