The paper presents the results of an experimental study on the mechanical behavior of a marly clay, motivated by the recurrent episodes of slope failure induced by intense rainfall events over the area of Langhe (NW Italy). The sensitivity of the material to environmental actions has been systematically explored by reproducing on laboratory samples hydrological and chemical processes similar to those expected to arise on site. Chemomechanical effects due to pore fluid dilution and decalcification have been investigated by testing both reconstituted and intact samples with different degrees of weathering. Pore water salinity was found to have minor consequences on the mechanical behavior of the material at its intact state and at the original calcite content. Decalcification enhanced the chemical sensitivity of the clay minerals constituting the soil matrix, not only by affecting the residual shear strength and the stiffness, but also by magnifying the tendency to swell under chemical loads, inducing a chemical sensitivity over swelling on saturation and enhancing water retention properties. Field observations also support the hypothesis that decalcification followed by interstitial pore water dilution are key factors to take into account to understand the degradation of the mechanical properties and that these phenomena can enhance the instability onset.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Landsliding in the Langhe area is a recurrent phenomenon that poses serious economic problems because of its frequency and the very high agricultural value of the land. Historically, major events studied occurred in Mondovì (1901), Levice (1927), Cissone (1941), Cigliè (1963), Arnulfi and Somano (1974), and Monastero Bormida (1994). Several others concurrent events occurred during November 1994 as a consequence of a catastrophic flood having recurrence intervals of centuries.
 The area, geologically known as Oligo Miocene Langhe succession, occupies the southern part of the Ligurian-Piedmont Tertiary basin and is characterized by terrigenous sedimentary rocks. Sedimentation occurred in a foredeep environment, with sand and silty marl strata alternating over thicknesses ranging from a few centimeters to a few meters (averages of half meter to one meter), deposited in a marine-continental environment. Tectonic forces and fluvial erosion gave to the area a hilly shape, typically with broad sides with strata parallel to slope dipping modestly toward NW and narrow sides at oblique angle to the slope dipping toward SE.
 Most landslides that have been recorded show a number of common features deserving a certain attention to understand the complexity of the mechanisms contributing to the instability processes. Instability developed generally as planar sliding in superficial marly clay strata that had recurrent characteristics in terms of grain size distribution and index properties (Table 1). Sliding planes have very modest dips (8° to 15°), mostly correspondent to the original bedding planes, and usually possessed a slickenside mirror-like aspect [Simeoni, 1998; Forlati et al., 1996]. Laboratory measurements returned ultimate friction angles of about 26°, that do not allow to easily justify the onset of sliding even when accounting for pore pressure distributions consistent with water flowing parallel to the slope [Chiappone, 1999]. Other tests showed that the material has the potentiality for developing important swelling pressures on soaking starting from low degrees of saturation and finally suggested a possible chemical sensitiveness in terms of swelling deformations upon pore fluid dilution [Forlati et al., 1998]. A most complete interpretation of the general frame from an engineering geology point of view will be provided in the work of G. Bottino et al. (manuscript in preparation).
Table 1. Statistical Analysis of the Grain Size Distribution, Calcite Content, and Index Properties of Specimens of Materials Interested by the Landslides of November 1994a
Grain Size Distribution, 339 Samples
Calcite Content, 118 Samples
Atterberg Limits, 324 Samples
Data are from Chiappone . Characteristic size d of particles: clay d < 2 μm, silt 2 μm < d < 60 μm, sand 60 μm < d < 2 mm.
Mean square deviation (%)
 The main mineral constituents of the marly clay layers are smectite, chlorite and illite (mass fractions ranging respectively from 15 to 30%, from 11 to 20% and from 10 to 15%) and calcite at varying contents (from 4% to above 40%). Structure is well developed at intact conditions: the apparent structural homogeneity is anyway often lost after unconstrained immersion of samples in water, when discontinuity surfaces with centimeter spacing open. Minerals of the discontinuity surfaces are the same of the intact matrix, but with a higher smectite fraction and a lower calcite one. As well, the destructured material covering the sliding surfaces has a higher methylene blue activity than the surrounding one in the intact matrix [Forlati et al., 1996; Chiappone, 1999; Veniale et al., 2002].
 Water circulation is associated with hydrogeochemical phenomena that include oxidation, water dilution and cycles of calcite dissolution-precipitation. Oxidation, extensively detected in situ, can be related to various processes, among which the most important one is perhaps pyrite oxidation. Indeed this reaction, observed in ESEM chambers by Veniale et al.  by inducing on Monastero Bormida clay several wetting-drying cycles, causes both acidification of the interstitial pore water and destructuration of the intact matrix, since pyrite is finally replaced by jarosite, that has a higher molar volume than the former. Conversely, dilution of interstitial pore water has been documented to occur through electrical measurements run in boreholes during the rainy days preceding the instability events of November 1994, while the increase in concentration due to water evaporation in the following days resulted in the deposition of fresh calcite concretions on the denuded sliding surfaces [Osella, 1995; Aiassa et al., 1996].
 Within this frame, the contribution of the present work concerns the evaluation of the influence of decalcification and pore water dilution on the mechanical behavior of Monastero Bormida clay. Hydraulic and chemical processes in line with those recognized in situ have been induced in laboratory on samples with different initial structures and calcite contents, outlining how these phenomena can influence the mechanical properties to such an extent that instability could be more easily conceived.
2. Monastero Bormida Landslide and Stratigraphy
 The Monastero Bormida landslide involved a soil mass of about 20 m width, 70 m length and 2 m height. It showed many of the typical features of the landslides of the area, since it developed quite quickly (three to four hours from initial movements to final collapse) and it occurred as planar sliding along the bedding plane of a marly clay layer. A few days after the event, the sliding surface appeared to be a mirror like slickenside plane, dipping about 12° (Figure 1).
Figure 2, taken from Osella , is a picture of the detaching surface, where the typical alternation of marly clay, cemented sand and marl was found. Strata thickness was in the range of 30–70 cm, with strong evidences of oxidation along preferential hydraulic paths in sandy layers, on the sliding surface and along lateral clayey joints. The oxidation stains would prove persisting acidic conditions deriving from the circulation of fresh water rich in CO2 and testify that at least a part of the detaching surface already constituted a preferential hydraulic path. Water content measured just below the surface was found to range between 12% and 14%, which in the hypothesis of full saturation would correspond to in situ void ratios ranging from e = 0.34 to 0.37. An intact block of material of about 1 m3 was collected just above the sliding surface soon after the landslide. In the following this material will be denoted as MB0.
 In September 2003 an inspection trench was dug to recover new samples from analogous strata below the original sliding surface, to be tested in the laboratory. The aim was reproducing a number of possible causes of mechanical degradation (loading-unloading cycles, drying-wetting cycles, salinity cycles) and possibly evidence how evolving hydrochemical conditions could affect the mechanical behavior and eventually contribute to instability.
 The stratigraphic succession found is given in Figure 3, together with the description of the main geological horizons and measurements of calcite content. The typical alternation of sandy and clayey layers was found again. Blocks of about 0.5 m3 were taken from the first intact clay level encountered (in the following denoted as MB1) and from the subsequent one (denoted as MB2). At the time of the retrieval, measured saturation degrees in the clayey layers were above 95%.
3.1. Basic Geotechnical Properties
 Index properties, specific surface and average calcite content of MB0, MB1 and MB2 provided in Table 2 are in good accordance with the statistical distribution previously reported and show a modest scatter between the three layers, mainly limited to calcite content. Unconstrained immersion of some blocks in distilled water baths induced the opening of discontinuity surfaces for MB0 and MB1 but not for MB2. Basic geotechnical properties of the material on those surfaces differed from those of the matrix for a slightly higher plasticity and a lower calcite content.
Table 2. Grain Size Distribution, Index Properties, Average Calcite Content, and Specific Surface of Monastero Bormida Specimens
 Specific surface Ss measurements presented were performed via methylene blue analyses, according to the French regulations [Afnor Norme Francaise, 1998] Total specific surface is here deduced through analysis of the chemical adsorption of methylene blue cations on the whole surface of minerals [Santamarina et al., 2002].
 MB0 and MB1 liquid limits wL were repeated at increasing NaCl concentrations, so to analyze the influence of physicochemical forces of interaction acting on clay particles. In fact, the liquid limit of chemically sensitive materials is known to drop exponentially at increasing ionic strengths [see, e.g., Di Maio, 1996]. Results as shown in Figure 4 show indeed a limited dependence.
3.2. Calcite Distribution and Pore Network
 A detailed inspection of the distribution of calcite at several scales was performed. First, its macroscopic content by weight fraction along the vertical direction of the 2003 trench was detected with a De Astis calcimeter, believed to offer more accurate measurements than other techniques [Lamas et al., 2005]. To verify the possible existence of heterogeneities that from a mechanical point of view could lead to localized weakness, measurements in the clay blocks covered extensively both the horizontal and the vertical directions, while single measurements were done in the sand and sandstone layers.
 In the first 1.50 m of altered clay alongside the trench, calcite content is below 10%. It tends to decrease in the upper sandstone layer, together with the encountered oxidations, while it is more abundant inside the MB1 layer, along which it increases from around 8% to 27% in a 35 cm space. This positive trend is orthogonal to the bedding planes, where distribution is very much homogeneous (Figure 5), and continues in the second sand layer (32% of CaCO3). By approaching the top of MB2 a steep drop is encountered, probably because of increased water circulation, enhancing the geochemical dissolution reactions at the bottom of the sandstone stratum and along the oxidation paths. Indeed a reversal, very important increase in calcite is found alongside the first centimeters of MB2, where in a 2 cm space calcite steps from 12% to 26%. In the following 12 cm increase occurs more softly from 26% to 40%, again orthogonally to the bedding planes (Figure 6). Data suggest that the current calcite distribution is associated with the geochemical processes and water paths of the site. Signs of oxidation, concomitant with acidification and normally associated to weathering, testify a more relevant chemical activity in the upper portions of both marly clay layers, that being in contact with the more permeable strata are more exposed to the evolving chemical conditions, so that the marked changes in calcite would be here associated to dissolution processes.
 Scanning Electron Microscopy (SEM) and X-ray map pictures of MB1 and MB2 gathered in Figure 7 confirmed that a good homogeneity persists at the microscopic scale, calcite being sparse over the soil matrix as nuclei of elongated ellipsoidal shape, with longer axis approximately of around ten microns. At a higher resolution (Figure 8), SEM pictures indicate that calcite is organized in grains with typical dimensions of a few microns spread all over the matrix. Its origin is both organic (coccholites and phoraminiphers) and inorganic (micrite crystals). In the detected structure, calcitic grains strengthen the soil matrix by physically bonding the clay aggregates: with respect to this aspect a more effective action could be attributed to micrite, a product of diagenesis processes in environments of high temperature and pressure, than to the fossil material deposited together with the soil mass. In MB1 specimens stains of calcite of detritic origin were found as well, testifying the occurrence of more recent cycles of dissolution and subsequent reprecipitation in a less crystalline form.
 The actual influence of calcite content over the pore network structure has been verified by means of Mercury Intrusion Porosimetry (MIP) tests on intact MB1 and MB2 specimens, as in Figure 9. Both materials have a mono modal pore size distribution, but the characteristic peaks at microstructural scale are anyway quite different (400 nm for MB1 against 90 nm for MB2). The shifting to larger values of the dominant pore size for MB1 can be related, among other factors, to the more open pore network developed as a consequence of a reduced carbonate fraction, which is consistent with the higher void ratio indicated Figure 9. In addition, the less constraint condition of the less calcitic sample also enhances the swelling tendency of the clayey fraction, which is also in correspondence with the larger dominant pore size detected on MB1. Collaterally, the higher void ratio of the less calcitic sample indicates that, at present and under low geostatic loads, calcite dissolution is not sufficient on its own to lead the material to instable conditions and therefore induce destructuration, as stated by the absence of a well developed macroporosity.
3.3. Mineralogy and Pore Water Chemistry
 X-ray diffraction (XRD) results (Figure 10) confirmed that the main mineral species are the same for the two layers. Beside calcite, quartz and some dolomite traces, the usual wide predominance of clay minerals was detected, with peaks in the spectra of smectite, illite, chlorite and kaolinite. Actually, the overlap between the mineralogic constituents of MB1, MB2 and MB0 as given in the literature is excellent. Indeed the collected information would suggest that the sequence of superficial marly clays of Monastero Bormida is characterized by an originally quite homogeneous material, now differentiating mainly because of the induced variations in carbonatic fraction.
 Pore fluid samples were extracted through squeezing from MB1 specimens [Iyer, 1990] and subsequently chemically analyzed. The chemical composition at intact and saturated condition has been extrapolated by conserving the mass of dissolved ions and assuming that at the involved concentration dilution does not alter the ratio between electrolytes. Accordingly, cationic species in decreasing order of concentration would be given by Ca2+ (613 mg/L), Mg2+ (138 mg/L), K+ (58 mg/L) and Na+ (33 mg/L), while anionic species are SO42− (1820 mg/L) and Cl− (40 mg/L). Dominance of calcium and of the sulfuric group is consistent with an environment where dissolution of calcite is occurring together with pyrite oxidation, concordantly with the oxidation stains of the site.
4. Experimental Program
 As introduced, laboratory experiments have been set up in order to evidence the mechanical sensitivity of Monastero Bormida marly clay to hydrochemical processes at the structural states that the material could experience in situ (intact, i.e., as in the matrix; or mechanically destructured). Water retention properties, volumetric behavior under oedometer conditions, swelling and collapse under load and shear strength properties were evaluated. Chemical processes considered were calcite dissolution and changes of pore fluid salinity (the latter by using NaCl and CaCl2 solutions as electrolytes). The three structural states tested were the intact one; the one obtained through mechanical destructuration induced by crushing and grinding of the intact material followed by remolding or compaction; the one obtained through mechanical destructuration and decalcification followed by remolding or compaction. Table 3 summarizes the testing program relevant to the present study.
Table 3. Testing Program of Concern for the Present Work
 Calcite dissolution has been induced in literature by means of several techniques. For instance, Griffiths et al.  adopted EDTA flushing to remove the cementation bonds of a structured soil loaded in an oedometer. In this study the reagents used were diluted HCl (chloridric acid) solution and CH3COOH (acetic acid).
 HCl dissolves calcite according to the reaction:
Since the effects of HCl on pH are dramatic, and this on its turn could affect to a significant extent clay mineralogy, adopted quantities of HCl were accurately evaluated according to the stoichiometric relationships of reaction (1) and calibrated on the actual quantity of calcite to be dissolved [Ligios, 2004]. After the treatment, soil powder was washed four times in distilled water to remove all the possible undesired reagents.
 XRD analyses were performed again on treated specimens (Figure 11) and compared with those of natural samples. The only difference found was the disappearance of the calcite peaks, so that the procedure was believed to be sufficiently mild to prevent the attack of clay minerals.
 Dissolution by means of acetic acid, through buffering with sodium acetate, invokes a less dramatic reaction since the pH of the suspension where the soil is immersed can be controlled and monitored during the treatment. By keeping the pH always around 5, neutrality with respect to clay minerals is expected to hold. Again, samples were washed four times after exposure to acetic acid to ensure the removal of reagents.
 Specific surfaces determined after decalcification imposed through the two different procedures are given in Table 4, where for ease of comparison results are gathered with the ones of the non decalcified samples. Data have been related to the pH of the soil suspension, since this is known to influence the Cation Exchange Capacity of amphoteric minerals. pH values were of about 7 for the untreated samples and for those treated with the HCl-distilled water baths procedure, of about 5 for the samples treated with CH3COOH and of about 2 for the specimens treated with HCl left unwashed. Indeed, in the latter case measurements through the methylene blue technique were not possible.
Table 4. Specific Surface of Monastero Bormida Specimens After Decalcification Through Different Methodsa
Dissolution With HCl and Four DW Baths
Dissolution With CH3COOH and Four DW Baths
Original Referred to Noncalcite Fraction Only
Dissolution With HCl and no DW Baths
Data are in units of m2/g.
 Both procedures induced a noticeable increase of specific surface, more relevant for MB2 that appears now to incorporate a slightly higher smectitic fraction than MB1. In the column “Original Referred to Noncalcite Fraction Only” of Table 4 the original specific surfaces of the natural materials are rescaled referring measurements only to their non calcite mass fraction. By comparing data of this column with those of the decalcified samples one can observe that calcite is actually “hiding” part of the clay specific surface (about 15 m2/g if the HCl procedure is taken as a reference).
 Grain size distribution changes due to decalcification were very moderate both for MB1 (26.5% clay fraction against an initial 25%) and practically null for MB2. Nevertheless, liquid limits wL were anyway increased, moderately in MB1 (from 46% to 49%) and more significantly in MB2 (passing from 40% to 52%). As a consequence, the plasticity index Ip of MB1 increased from 17% to 20% and the one of MB2 from 14% to 22%.
5. Water Retention and Water Permeability Properties
 The water retention properties of the MB1 material at different structural states and during a wetting path under unstressed conditions are presented in this section. The understanding of the water retention properties (hygroscopicity and water storage capacity) is an essential element for the comprehension of the swelling capabilities and the water sensitivity of the clayey fraction of the material as it undergoes the different structural states described in the previous section.
 The wetting paths started on the specimens equilibrated at a relative humidity of 75% using vapor equalization technique. Transistor psychrometers [Woodburn et al., 1993] with an extended measuring range were used to obtain the retention curves in the total suction range from 40 MPa to 1 MPa. The output voltage of the transistors was related to the relative humidity (or total suction via the psychrometric law) through a suitable and extended calibration [Mata et al., 2002]. The specimens were wetted in steps by adding distilled water drops on the material, stored for 1 day equalization, weighed and the relative humidity of the air surrounding the soil measured. At the end of the multistage wetting procedure, the specimens were weighed, oven-dried and the water contents back-calculated. Measurements were performed on two specimens for each structural state.
 Water retention results are shown in Figure 12, jointly with the retention curve of intact MB2 material obtained from MIP results. In this last case, the injection of mercury has been admitted to be equivalent to the ejection of water for the same diameter of pores being intruded, as suggested by different authors [Prapaharan et al., 1985; Romero et al., 1999]. As observed in Figure 12, the intact material displayed a reduced water storage capacity at low suctions (<1 MPa) as a consequence of the relatively low void ratio of this structural state. The results at high suctions (>7 MPa) for the same structural state tend to a linear relationship between the logarithm of suction and the water content. Romero and Vaunat  indicated that at low water contents (zone of hygroscopicity or water adsorption), where the influence of void ratio is negligible, the suction–water content relationship is mainly dependent on the specific surface of the clayey fraction.
 On mechanical destructuration, the material is capable of retaining slightly more water in the low suction range (<1 MPa). Despite some shifting toward somewhat lower water contents in the zone of hygroscopicity, this mechanically destructurated state presents an equivalent slope of the linear log suction–water content relationship, which suggests that the specific surface has not changed significantly. On decalcification, a different water retention behavior is clearly detected on the less constraint clayey fraction. For one, the water storage capacity in the low suction range (around 1 MPa) is drastically enhanced. For another, the linear log suction–water content relationship in the high suction range displays a flatter slope, which clearly denotes a larger specific surface at this decalcified state [Romero and Vaunat, 2000]. The higher specific surface, which is in agreement with data presented in Table 4, is associated with a higher hygroscopicity (specific surface able to adsorb water) and a higher capacity of undergoing expansive volume changes on wetting.
 A complementary hydraulic study on the water permeability properties under saturated conditions was carried out, making use of the oedometer tests on intact and remolded samples described in the next sections. Time evolution of soil deformation in these tests has been interpreted to determine the different conventional parameters used in the consolidation analysis following Terzaghi's 1D theory.
Figure 13 presents the variation of the saturated water permeability as a function of void ratio for intact MB0 and remolded samples of MB1. Remolded samples at their natural calcite content and decalcified states were prepared by mixing the soil dry powder with distilled water and 1M NaCl solution at 1.2 times wL. As detected in Figure 13, the saturated water permeability of the intact material is consistently below the values of the remolded material for the same void ratio. This fact is a consequence of the different pore size distributions associated with the different structural states. The monomodal pore size distributions of the remolded states are expected to present dominant pore sizes much bigger than the ones presented in Figure 10. In addition, the permeability of the decalcified state is also regularly maintained below the values of the remolded samples at their natural calcite content and for equivalent void ratios, fact that is not easily explained unless pore size distribution changes are considered. With respect to pore water dilution aspects in the remolded state, it is clearly observed that permeability increases with salinity in both structural states (at their natural calcite content and decalcified ones). This aspect can be ascribed to the chemical sensitivity of the clayey fraction of the material and the associated pore size distribution changes. When imbibed with saline solution, the microstructure tends to aggregate, usually explained in terms of the depletion of diffuse double layers, increasing the macroporosity between aggregates, and thus the water permeability. In fact, the increase in permeability with salinity for high-salinity permeants is more evident when Na+ is the predominant cation in the solution [Studds et al., 1998; Pusch, 2001; Villar et al., 2005; Villar, 2006; Castellanos et al., 2006].
6. Volume Change Behavior Under Oedometer Conditions
6.1. Compression Behavior of Intact Samples
 An insight into the effectiveness of structural bonding in the natural intact material and on how mechanical actions could ultimately lead it to destructuration can be pursued by analyzing its response under high-load oedometer tests, such as those reported by Osella .
 Intact MB0 samples were loaded up to maximum vertical effective stresses of 50 MPa (Ismes samples, from the laboratory where they were carried out) and 25 MPa (Enel Cris samples). Those results are gathered together with the intrinsic compression line (ICL), the sedimentation compression line (SCL) and the intrinsic swelling curve (ISC) in the Iv − log σ′v plane in Figure 14, as according to the normalization process suggested by Burland . Here Iv is the void index (Iv = (e − e*100)/(e*100 − e*1000)), where e*100 and e*1000 are the void ratios of the remolded clay under vertical effective stresses σ′v = 100 and 1000 kPa respectively. In Figure 14, the ICL and the ISC proceed from experimental data of an oedometer test on remolded MB0, commented later on, while the SCL has been drawn on basis of the best fit regression line proposed by Burland .
 Loads higher than 20 MPa are needed to cross the ICL, proving the persistence of an important structure. Only from a vertical effective stress of 20 MPa onward the compression curves begin to bend downward, which is a sign of initiation of a progressive structural breakdown [e.g., Burland et al., 1996].
 As for deformability upon loading, it is possible to identify roughly three zones, depending on the applied load (Table 5). Below 3 MPa the samples have a rather sparse compressibility; between 3 and 20 MPa the reload index Cr ranges quite uniformly around 0.061; above 20 MPa the compressibility index Cc rises quite markedly, with a renewed scatter between different tests, but remains still well below the intrinsic compressibility index Cc* = 0.321. This latter aspect would suggest that vertical stresses higher than 50 MPa are needed to ensure full destructuration.
Table 5. Oedometer Compressibility of Intact MB0 and MB1 Samples
Cr (0–3 MPa)
Cr (3–20 MPa)
Cc (above 20 MPa)
MB0 – Ismes A
MB0 – Ismes B
MB0 – Ismes C
MB0 – Enel Cris 1
MB0 – Enel Cris 2
MB1 – DW
MB1 – 1M NaCl
 Swell sensitivity values C*s/Cs [Schmertmann, 1969], where C*s and Cs are the swelling indexes for the remolded and the intact samples respectively, are lower than 1.25 for three samples over five. Such low values, counterbalanced by other two quite higher ratios, indicate that the loading-unloading procedure up to the stresses imposed can, but not always, cause debonding of the microstructure leading the material toward a destructurated state.
 Influence of the pore water salinity over the oedometer compressibility of the intact material was verified by applying two loading-unloading cycles over a MB1 sample that was initially saturated with distilled water and then with a saline solution (Figure 15). The oedometer used was designed on purpose, so to ensure full saturation and control of the chemical composition of the pore fluid by means of back pressure applied from the top and bottom caps [Chighini, 2006]. During the first mechanical cycle with distilled water a behavior analogous to the one of MB0 was recorded. After unloading a unity hydraulic gradient was imposed and the pore fluid replaced with a 1 M NaCl solution. Flushing was continued until the electrical conductivity of the solution leaving the sample was equal to the injected one, then another loading cycle was imposed. A modest void ratio reduction was detected upon salinity increase under the small vertical stress applied while no noteworthy chemical effects upon compressibility were appreciated during the mechanical cycle pursued with saline solutions. Consistently, the void ratio achieved upon unloading was indeed the same as the one of the first cycle with distilled water, so that on the overall the oedometer behavior of the intact material appears as substantially independent on the salinity of the pore fluid.
6.2. Compression Behavior of Remolded Samples
 Remolded samples of MB0, MB1 and MB2 at their natural calcite content were prepared by mixing the dry soil powder with distilled water and with a 1M NaCl solution at 1.2 times wL. Table 6 compares the compressibility of the remolded samples under the various chemical conditions explored.
Table 6. Oedometer Compressibility of Remolded Samples Under Different Chemical Conditions
MB0 – DW
MB1 – DW
MB1 – NaCl
MB1 – DW after diffusion
MB2 – DW
MB2 – NaCl
MB2 – DW after diffusion
 In both the saline and the distilled water environments, MB2 originated curves laying below the other ones (Figure 16) and a somewhat stiffer behavior during virgin loading, as a reflection of the smaller liquid limit related to the higher calcite content. Also in this state, the effect of pore water chemistry on virgin compressibility appears ambiguous, MB1 displaying a moderate stiffness increase with salinity and MB2 a contained reduction.
 The samples originally prepared with saline solutions were exposed to free diffusion of distilled water after a complete loading-unloading cycle, under a vertical effective stress of 50 kPa (Figure 17). Dilution of the pore fluid caused some limited chemical swelling for both MB1 and MB2 (strains around 3%). These results will be commented within the light of a more general interpretation of the swelling under load behavior presented in the next section. In the following mechanical cycles, only minor consequences that could be ascribed to the changes of the chemical environment were detected.
 The same experimental procedure was adopted for samples prepared after decalcification by means of the HCl procedure (Figure 18). In this respect, MB2 experimental points lay above the MB1 ones, manifesting finally the consequences of its greater clay content and higher specific surface at this state. Compressibility is consistently increased in all samples, but particularly for MB2, where the swelling coefficient in all conditions is about doubled (Table 6). The most noticeable aspect rests anyway on the fact that the chemical swelling due to distilled water diffusion is significantly magnified, being of 9% for MB1 and of 12% for MB2 (Figure 19). Congruently with the effects detected on permeability, the times needed to complete the diffusion process and the associated chemically induced swelling are larger for the decalcified soils.
 Intrinsic reload and swelling indexes of compressibility C*r and C*s after distilled water diffusion were always higher not just than the corresponding ones for saline conditions, but as well than those detected in samples prepared with distilled water since the beginning. The effect is once again more evident for the decalcified material. That would be ascribed to the replacement of Ca2+ ions in the mineralogic structure of the smectite with the Na+ ones of the saturating solution. It is well known in fact that [e.g.,Mitchell, 1993], when tested at the same electrolyte concentration (in the current case distilled water), sodium smectite is more compressible than the analogous calcium one. As proved elsewhere in the literature [e.g., Sridharan and Jayadeva, 1982; Di Maio, 1996] saturation with a concentrated solution can induce cation replacement inside of the mineral structure and consequently affect its compressibility. It is believed that in the present tests an analogous phenomenon took place: in this respect the more pronounced effect expressed by the decalcified samples could be due to the fact that calcite hinders the response both because of a mechanical constraint (contributing to a stiffer material) and because of a buffering action on the replacement of Ca2+ cations.
 It can be of some interest to notice that, considering all the three materials and the saturating solutions used, a good match is found between intrinsic compressibility index C*c and the void ratio at the liquid limit eL (Table 6 and Figure 20), under the relationship C*c = 0.356 eL − 0.10, relatively above the trend provided by Burland  (C*c = 0.256 eL − 0.04).
 Swelling tests on intact MB0 samples presented by Osella  have been compared with other tests run on MB1 samples statically compacted in the laboratory. The procedure followed was to monitor in time the swelling or collapse response on soaking under constant load starting from hygroscopic conditions and initial water contents of about 2%. Vertical stresses imposed ranged from 50 kPa to 800 kPa. Swelling pressure values have been extrapolated by plotting the results in a (σv, ɛ) plane, σv being the constant vertical stress imposed alongside the test and ɛ the final deformation developed during hydration.
 The destructured samples used for comparison were prepared in the laboratory starting from the MB1 material both at the natural calcite content and decalcified. Because of limitations of the compaction process, the compacted samples possessed void ratios of e = 0.47 and e = 0.67, in front of the smaller value of e = 0.37 for the intact ones. These void ratios refer to samples including calcite. For the decalcified samples it was chosen to preserve the density of the non calcite fraction which ended in actual void ratios of e = 0.63 and e = 0.95. For ease of comparison, those decalcified tests will be in the following labeled as e* = 0.47 and e* = 0.67. Indeed, swelling behavior is expected to depend both on the overall density and on the density of the clay-active minerals. With respect to the problem under study, preservation of the density of the non calcite fraction was believed to offer more information in conjunction with the processes as evolving in situ. Nevertheless a comparison in terms of total void ratio is possible when looking at the results of the compacted samples of e = 0.67 and decalcified compacted e* = 0.47 (e = 0.63). Influence of pore fluid chemistry was tested by saturating both with distilled water and with a 1 M NaCl solution.
 The role of the structure and of the mechanical integrity of the calcite bonds is addressed in Figure 21. Results emphasize the fact that the calcite bonds present in the material at its intact state actively concur to constrain the material, limiting the swelling strains developing under small vertical stresses. The swelling pressure is around 800 kPa for the intact material, somewhat higher for the material compacted at e = 0.47 and drops to 350 kPa for the material compacted at e = 0.67. Since the process is magnified by higher densities, it is possible to imagine that the swelling pressures of samples compacted at the same void ratio of the intact material would be quite higher, probably comparable to the 1.8–2 MPa measured by Forlati et al.  on MB0 intact samples retrieved in the very proximity of the sliding surface. On this basis, it is assumed that those very sustained values have to be related to a material that has already experienced some kind of destructuration, for instance induced by mechanical processes such as the action of tectonic forces.
 As for the compressibility of non decalcified samples, the role of salinity appears to be questionable or at least limited, as can be appreciated noting that no specific trends relating the pore fluid adopted to the induced deformations have been found.
 In Figure 22 results of swelling under load tests on the decalcified samples are reported together with the trends obtained for the non decalcified samples to ease the comparison. Given the entity of the deformations developed, attention must be paid to the fact that the scale has been changed with respect to Figure 21. At low stresses, decalcification enhances swelling, whereas an increase in collapsibility is detected at high stresses. As a consequence of these opposite effects, the swelling pressure reduces with the decalcification process.
 Although not dramatically, the role of salinity is here more evident than before, since water dilution consistently leads to higher strains both in swelling and collapse conditions. This aspect could be ascribed to the consequences of the enhanced chemomechanical sensitivity of the material in absence of calcite as seen in the previous section (see, e.g., Figure 19 and Table 6). In the low stress range, when swelling is dominating, the dilution of the salinity enhances repulsion of the clay particles inducing higher strains, in line with what observed for chemical swelling. On the contrary, when collapse dominates, the dilution of water salinity reduces the stiffness of the samples. The collected evidences suggest, as a working hypothesis, that decalcification leads the geometric fabric to be more sensitive to pore water chemical changes, both from a qualitative and a quantitative point of view.
7. Shear Strength
 Three direct shear (DS) tests were done during the present study on MB1 specimens possessing different calcite contents of 9%, 18.5% and 23.4% (Figure 23). To ensure complete saturation, samples were first loaded, back pressured, flushed with distilled water and then sheared along the direction of the bedding planes. As for analogous tests on MB0 presented by Chiappone , peak shear strengths values are quite high and dispersed and at present do not allow for any sound interpretation of a peak failure envelope. Indeed, as reported in the existing literature [e.g., Del Olmo et al., 1996], calcite bonds influence noticeably the peak strength of natural clays, so that with respect to this aspect Figure 23 simply proofs further that a correct interpretation shall be done at homogeneous calcite contents. The same shall hold for the ultimate shear strengths τult, recovered after nine shearing cycles, although results appear here to be relatively more comparable, so that an hypothetical envelope with c′ult = 25 kPa and ϕ′ult = 13° could be sketched.
 The intact samples failed after a displacement of a few tenths of mm with a following dramatic loss of strength. In Figure 24 the shear stress-displacement curves of intact MB1 sheared under vertical effective stresses σ′v = 50 kPa and σ′v = 200 kPa are given (the plot is limited to the first two shearing cycles in order to appreciate the peak reached at very modest displacements, as typical of well structured soft rocks).
 By referring the strength loss to the peak strength according to the index of brittleness Ib, defined by Bishop  as:
and by cross checking it with calcite content and applied stress (Figure 25) it is made evident that the brittleness of the intact material is relatively independent of the stress level, while a more sounding relationship is found with respect to calcite content.
 By conducting multi stage Ring Shear (RS) tests [Bromhead, 1979] on remolded MB1 and MB2 samples and DS tests on MB1 large blocks (30 × 30 cm), where displacements up to 20 mm were imposed along the discontinuity surfaces opened after unconstrained immersion in water, a comparison between the ultimate strength available along the bedding planes and the residual strength of the intact material has been done.
 Calcite content along the discontinuity of the block samples tested was on the average of 14%; the consequent failure envelope reported in Figure 26 is slightly curved, accordingly to Skempton . Following the approach proposed by this author and indicating as residual friction angle ϕ′r the one characterizing the secant slope joining the origin of the (σ′, τ) plane to the experimental points obtained under a vertical effective stress σ′v = 100 kPa, it is here found that ϕ′r = 21°.
 Residual RS strengths of remolded samples proceeding from MB1 (CaCO3 = 15%) and MB2 (CaCO3 = 34%) were comparable to those of the block samples, since the residual angle of shearing resistance ϕ′r found is equal to 19.1° for MB1 and to 22° for MB2. As in the literature [Alonso, 1998] the resistance measured for the same material (MB1) along the cleavage surface is higher than the one returned under ring shear conditions.
 Wetting-drying cycles, such as those that could occur because of seasonal water fluctuations, do not affect residual strength, as proved by a RS test run on a MB1 sample previously subjected to three wetting-drying cycles. This can be appreciated in Figure 26, where no differences are tracked with respect to the other RS tests.
 At the natural calcite content a limited influence of salinity is detected, as remarked by Figure 27 where the envelopes for 1M and 2M NaCl and 1 M CaCl2 saturated samples are given. In terms of ϕ′r, the maximum increase is limited to 1°.
 Shear loss due to decalcification (Figure 28) was found to affect strength so much to reduce it to values comparable with the actual in situ slope of the bedding planes (ϕ′r = 14.5 ° for MB1 and ϕ′r = 12° for MB2). Here the more marked drop of strength shown by MB2 would be related to its higher specific surface. In decalcified condition consequences of salinity increase on shear strength are finally well more appreciated, since a 1M NaCl recovers MB1 friction almost to the same value of non decalcified material (ϕ′r = 18°).
8. Considerations on the Effects of Mechanical and Chemical Actions on Shear Strength
 As discussed, bonds existing between particles at the intact state are responsible for very high peak strengths and pre failure shear stiffness. Once that shearing overwhelms cementation a post peak behavior is onset, where as pointed out for instance by Alonso and Gens  two different mechanisms are easy to recognize. First, strength is dramatically reduced by the destruction of cementation along the induced failure surface. Then a smoother decline toward residual values occurs because of geometrical rearrangement of loose particles, as common in soils. The first mechanism, pretty much dominant for Monastero Bormida marly clay, depends mainly on the calcite content, while no significant trends are evidenced with respect to the normal stress applied (Figure 25).
 A coherent unique vision of shear behavior is not possible considering just the results of DS tests. Indeed, ultimate strengths obtained this way are still quite higher (order of 6°) than those returned by testing the surfaces of discontinuity or by RS tests, as a consequence of the undulated failure surface developed.
 Analysis of shear displacement curves of these tests during first shearing under a vertical effective stress of 50 kPa, for convenience of discussion here drawn in a non dimensional frame with respect to the peak shear (Figure 29), outlines that remolded samples show a post peak behavior which is not evidenced along the sheared surface. In the mechanically destructured state this post peak behavior can be referred to stress level and load history. At this state, calcite does not appear to influence strength loss in the same manner as in the intact state, since when passing from CaCO3 = 15% (MB1) to 34% (MB2) results are almost identical. On the contrary, complete calcite removal induces greater strength losses, related to the “effective” shape of particles involved in the shearing mechanism. Indeed, calcite particles as evidenced in the microscopic study are cubic or spheroidal, shapes that do not allow manifesting the potentiality for preferential orientation upon shearing. The latter is instead favored by the platy shape of clay minerals, so that a sharper transition toward a complete oriented shearing surface is possible in the case of the decalcified samples.
 As far as the reduction of residual shear strength upon decalcification is concerned, in a well documented paper, Hawkins and McDonald  evidenced that, for Fuller's Earth marly clay, calcite increased residual strength. By comparing strength, calcite content and Ip of natural samples with different carbonate content and of samples progressively decalcified, they found that the residual friction decreased quite linearly from about 30° to about 6–8° when calcite passed from 60% to 28%. At contents lower then 28% further decalcification appeared to have no consequences, so that this value represented a clear threshold between transitional and sliding mode for particles involved in the shearing process [Lupini et al., 1981]. According to this concept, the residual angle developed in a purely sliding mode is equal to the coefficient of interparticle friction, and therefore would depend only on the mineralogy in pure homogeneous clays; while in the transitional mode geometrical effects due to the presence of rounded particles increase residual friction. Lupini et al.  express dependency on interparticle geometrical effects in terms of the plasticity index Ip. By doing so a threshold values for Fuller's Earth was found at about Ip = 40%.
 In Figure 30 the results of the present study are interpreted jointly with those of Hawkins and McDonald , by making reference both to calcite content and to Ip. Although the population is here limited, the transitional behavior is evident in both interpretations. It has to be remarked that even full decalcification does not onset the sliding mode for Monastero Bormida clay, since no lower plateau is reached. Interestingly, the plot in terms of Ip offers a better match. It appears that the plasticity index is more able to gather together the effects of calcite and clay content, effectively combining them in a unique variable. It is as well of interest to point out that the good trend encountered in terms of Ip accomplishes for data proceeding from both MB1 and MB2, reflecting both the evolution of decalcification induced in situ conditions and in the laboratory.
 On their whole, these comparisons show that the residual shear strength of Monastero Bormida clay is governed both by calcite content and activity of the clay fraction. At the original calcite content, even when modest with respect to similar literature data, calcite greatly hinders the intimate nature of clay minerals. The latter is brought completely to light only by decalcification, as reflected by the increased sensitivity to pore water salt concentration, in line with literature as for clays containing a significant smectitic fraction [e.g., Moore, 1991; Di Maio and Fenelli, 1994].
9. Summary and Conclusions
 Experimental tests performed during this and previous studies show many evidences of a complex mechanical behavior for the Monastero Bormida marly clay, on which different orders of structure exert their effects. The volumetric behavior of the intact material exposed to loading-unloading cycles at high stresses shows, even if not consistently, potentiality for the occurrence of mechanical debonding and destructuration. Local calcite content measurements indicate steep changes, but never well marked sharp discontinuity; while on the other hand SEM analyses prove that calcite is present both under organic and inorganic forms. It seems reasonable to postulate that the micritic inorganic fraction, that formed as a product of diagenesis under high loads and temperatures and has a crystalline structure, might be more effective in contributing actively to structure, and on the other hand that the portions of intact material more prone to destructurate under mechanical actions are those less rich in micrite and not in calcite fraction as a whole.
 Even grinded, crushed and afterward reconstituted, the mechanically destructured material appears to be only modestly sensitive to salinity or saturation effects. This could sound unexpected, since XRD analyses identify smectite as the dominant mineral of the clay fraction. Indeed, even in the disaggregated state and at moderate contents, calcite still offers a decisive role in governing mechanical properties. This role appears to be important with respect to all of the main hydromechanical aspects, such as oedometer compressibility and residual shear strength, or swelling under saturation and salinity changes and water retention properties, still typical of a material with little sensitivity to chemical effects.
 Dissolution of calcite reveals the intimate behavior of the clay minerals composing the soil matrix. Decalcification induces a significant drop of the residual shear strength, consistently with the increase in plasticity index along a transitional sliding mode as described by Lupini et al. . Salinity effects, not appreciated originally, are now quite consistent, so that water dilution becomes a parameter influencing shear strength. As well, all the other mechanical features appear at this point in line with those of a water chemistry sensitive material. Swelling is sensitive to saturation as well as to salinity and the compressibility index decreases mildly with increase in salinity.
 In general, calcite appears to offer at various states and in various forms a clear contribution. At intact conditions provides bonds that effectively increase the stiffness and actively concur to increase the shear peak strength, reflected afterward in a higher brittleness. After mechanical destructuration, geometrical effects still contribute to the residual shear strength and sustain the stiffness in oedometer conditions. As well, at this state the physicochemical interaction between the pore fluid and the clay minerals is still masked, as testified by the very modest coupled effects recorded during swelling, shear strength and oedometer compression tests. Mechanical properties are then furthermore degraded upon calcite dissolution. Here, besides experiencing a significant shear strength reduction, an increased compressibility and a higher water retention capability, the clay is free to express its intrinsic chemo mechanical coupling, as evidenced by the effects measured in terms of property changes induced by different saline solutions.
 Therefore, according to the results of the study, a sequence of processes including first mechanical destructuration, followed by calcite dissolution and pore water dilution would pose the basis for more critical mechanical conditions. Such a sequence of events, in line with commented in situ evidences, could aid to conceptually conceive the onset of landsliding.
effective cohesion intercept at ultimate conditions, kPa
compression index, unitless
intrinsic compression index, unitless
reload index, unitless
swelling index, unitless
intrinsic swelling index, unitless
void ratio, unitless
void ratio at liquid limit conditions, unitless
void ratio of remolded samples in oedometer conditions under a vertical effective stress σ′v = 100 kPa, unitless
void ratio of remolded samples in oedometer conditions under a vertical effective stress σ′v = 1000 kPa, unitless
plasticity index, %
void index, unitless
specific surface, m2/g
liquid limit, %
plastic limit, %
vertical strain, %
residual angle of shearing resistance, °
ultimate angle of shearing resistance, °
vertical stress, kPa
vertical effective stress, kPa
normal effective stress, kPa
shear stress, kPa
shear stress at peak conditions, kPa
shear stress at residual conditions, kPa
shear stress at ultimate conditions, kPa
 The authors wish to thank Ivan Berdugo for performing water retention and reconstituted oedometer tests on MB0 samples, Analice Lima for performing MIP on MB1 and MB2 samples, Gianclaudio Ligios for the oedometer analyses on reconstituted MB1 and MB2 samples and for calcite distribution characterization, and Cristina Jommi for aid in reviewing the manuscript. Special thanks are due to Renato Lancellotta, who tutored Serena Chighini during her doctorate, for giving impulse to the study, for the constructive discussions, and the participation to the work. Traveling expenses have been covered through the integrated action Piemonte-Catalunya: Ajuts per a accions de cooperació i mobilitat de la Direcció General de Recerca de la Generalitat de Catalunya (ACI-2003-35).