The samples were prepared by cutting intact sample into small cubes (0.5 × 0.5 × 0.5 cm). These small samples were immediately freezing-dried. Undisturbed fragments produced by breaking carefully the dried samples were directly analyzed for their microstructure by using a scanning electron microscope (SEM). Furthermore, elemental compositions of the samples corresponding to the SEM imaging were also identified by using energy dispersive X-ray analysis (EDS) analysis.

Specific surface areas (SSA) of marine sediments were determined using nitrogen adsorption based on B.E.T method. Total Cation exchange capacity (CEC_tot) and cation exchange capacity for each exchangeable cation (Na⁺, Ca²⁺, K⁺) of these sediments were determined using Metson method according to the NF X 31-108 standard.

The samples were prepared by packing about 50 mg of the marine sediments in a platinum crucible. The prepared samples were immediately placed in the thermal balance. After 10 min of the balance stabilization, these samples were gradually heated from 20 °C to 1100 °C with a linear speed of heating (10 °C/min). According to the mass losses and the corresponding thermal spectra, we can quantitatively determine the mineral compositions of sediments.

The sequential extraction tests were conducted with marine sediments and distilled water. Three identical samples were prepared by mixing 9 g of the natural sediment in 30 ml of distilled water with initial pH value of 6.3. The prepared samples were gently agitated for 48 h. And then, the samples were centrifuged for 30 min at 2450 g to separate the solid from the dissolved phases. The suspensions were subsequently filtrated with filters of 4.5 μm pore size. Finally, sodium and calcium concentrations in the dissolved phase were analyzed using an atomic adsorption spectrophotometer (Shimadzu AA-6300). Following the first extraction, the tested samples were dried in oven under temperature of 40 °C for 24 h. These dried samples were used to prepare the second set of samples according the first extraction procedure. The third extraction tests were sequentially conducted. The coefficient (D) of cation desorption capacity is defined and calculated as follows:

The undrained cohesion was measured by the fall cone method (cohesion denoted c_uf) illustrated by squares in Fig. 1c, and the laboratory vane test (cohesion denoted c_uS) illustrated by circles. Their profiles are compared with Skempton's equation (3), expressed for remoulded normally consolidated mineral clays.

The results show that the values of c_uf and c_ul are close to each other in the first 9 metres, c_uf being slightly lower from a 9 m depth. Compared to Skempton's correlation, the measures are slightly higher in the first 7 m. This is due to the presence of natural cementation which can be very high as shown between 7 and 11 m, where the measures move strongly away from the correlation.

The mechanical behaviour is illustrated by three oedometric tests (Fig. 3), performed on the GoG sediments in the following conditions:

• • GoG_Nat (5_10 11): specimen of intact sediment taken at 10.8 m depth;
• • GoG_Nat (2_11 12): specimen of intact sediment taken at 11.3 m depth;
• • GoG_REM: specimen of remoulded sediment.

In this test, the applied stress tensor to the material is such that we impose the evolution of both parts of the stress, the isotropic path and the deviatoric one. The deformation obtained depends only on the amplitude of the stress tensor on the case of the remoulded sediments (thus, two mechanisms of deformation, isotropic and deviatoric), in which it is necessary to add the effect of bonding damage in the case of the intact sediment (Hattab and Favre, 2010; Yin et al., 2010).

The GoG_REM curve (Fig. 3), were e is the void ratio and σ’_v the vertical effective stress, exhibits the intrinsic compressibility phenomenon of the sediment (Burland, 1990; Favre and Hattab, 2008) characterized by the compression index C_c = 0.94.

The observed paths, for both intact sediments, highlight the behaviour of a non-remoulded (intact) normally consolidated sediment, with a bend towards the stress σ’_v = 60 kPa located above the remoulded oedometric path, suggesting the beginning of breakage of cementation between “grains” and particles. That also shows the elastic limit in the phenomenological behaviour of the material. This bend is followed by a curve which seems to gradually join the remoulded behaviour, expressing a process of progressive degradation of the cementation between the “grains” and plastic flow rule. This is clearly shown in the case of GoG_Nat (2_11 12) sample. GoG_Nat (5_10 11) sample seems to contain higher cementation (more sensitive sediment (Skempton and Northey, 1952)).

In the aim to observe the local deformation, especially which bond is concerned by the damage, Mercury porosimetry intrusion tests were performed, supported by SEM pictures. These latter tests cannot otherwise determine the nature of the damaged bond but only their location.

The samples of intact sediment were extracted from the core, the overburden effective stress being σ’_v0 = 36.5 kPa. The porosimetric curves in Fig. 4b show a very high porosity of the natural sediment, which corroborates the behaviour obtained at the macroscopic level (Fig. 3). The pore size distribution of the sediment is complex and the curve presents several peaks in the 15 to 40 μm range which may correspond to inter or trans-assemblage pores according to Collins and McGown (1974) terminology. One observes the presence of two peaks in the curves (Fig. 4a), with a first pore family which may be considered as an interaggregate type, with an average diameter of about 1 μm; and a second pore family of the intra-aggregate type with an identical average diameter of about 0.07 μm. After the oedometric loading, the porosity change seems to be directly related to the closure of the interaggregate pores, and then, the breakage of the cementation which link these aggregates. This is clearly highlighted by the kinetic of the incremental curves from the initial state to the oedometric loading state (Figs. 3 and 4a) where we can see:

• • at σ’_v = 200 kPa, the average diameter of the interaggregate pores decreases significantly, varying from 1 to 0.4 μm in the intact sediment. The intra-aggregate porosity remains unchanged, with an average diameter of about 0.07 μm;
• • at σ’_v = 1 MPa, the average diameter of the interaggregate pores continues to decrease, with a value equal to 0.25 μm.

SEM observations (Fig. 5) after the oedometric loading of σ’_v = 1000 kPa corroborate MIP results. The photo of the remoulded sediment (Fig. 5a) shows particles clearly oriented following the horizontal plane (which is perpendicular to the loading axis). Particles in this case are in fact free to rotate. On the other side, the aggregates structures in intact sediment are not affected by the mechanical loading (Fig. 5b). Particles inside the aggregate remain bonded by a cementation which is not damaged yet.

The interest in what follow is in trying to understand the nature of the glue concerned by the degradation; several complementary physicochemical tests were thus conducted, starting with SEM picture analysis.

The SEM photo of intact sample (sample Sed 2, see Table 1) in Fig. 6 showed that the natural sediment presents a typical microstructure similar to card-house. It is also confirmed that the relative high void ratio of intact samples compared to the remoulded one results in large interaggregate pores. Furthermore, as it was proven above, the bonding exists between sediment aggregates. It is supposed that these bonds are out of a smectite gel according to their fine structure.

The results of EDS analysis in Table 2 indicate that main chemical elements are Na, Mg, Al, Si, S, Cl, O, K, Ca, Fe in tested sediment, and the mass proportion of sodium and calcium present the highest values. So it can be suggested that the interlayer exchangeable cations are mainly dominated by Na⁺, Mg²⁺, K⁺, Ca²⁺, and Fe²⁺ as other compensating cations.

Table 3 presents the surface area (BET) and C.E.C results for the sediments and two quasi pure clays (Speswhite Kaolinite and Wyoming Bentonite). It can be noted that the surface area (BET) values for the sediments are close to that of Wyoming Bentonite except for Sed-1. This highlights that the particle sizes of the tested sediments are similar to that of smectite.

The C.E.C results show that the C.E.C values for Na⁺, Ca²⁺ and K⁺ of the tested sediment are higher than that of Wyoming bentonite, and its C.E.C value is higher for Na⁺ compared to that of the other two cations. In other words, the smectite present in the sediment is sodium one, but the exchangeable calcium content is very close to that of sodium. In addition, adding the C.E.C. for three exchangeable cations leads to an unusually high value, which is largely greater than the proposed total C.EC. This can be explained by the testing procedure: the total C.E.C value is determined using ammonium ion while those for the exchangeable cations by measuring their concentrations of first extraction solution after equilibrium of ammonium ion-sediment slurry. This indicates that the sediment contains also large amount of salts in pore water. From the sodium and calcium abundance of the tested sediment, it can be assumed that this marine sediment is very sensitive to the changes of pore water chemistry.

The mineral compositions of the sediments estimated by thermal analysis are presented in the Table 4. The results show that these sediments contain mainly kaolinite and smectite. The composition comparison between the sediments with different depth indicates that these sediments are relatively homogenous except for Sed 1. These results are consistent with that aforementioned obtained by using X-ray diffraction analysis.

Table 5 presents the cations (Ca²⁺ and Na⁺) desorption capacity of sediments for three extractions. During the extractions, interstitial saline solutions are firstly diluted in the suspension, precipitated salts are subsequently dissolved, and cations fixed on clay particle surfaces are finally released. In this study, we consider that the first two effects mainly occur in the first extraction, and the last effect predominantly in second and third extractions. The results show that the desorption capacity proportions between Na⁺ and Ca²⁺ for the tested sediments vary from 5.5 to 15 for the first extraction, and from 6.0 to 19 for the second and third extractions. This indicates that Na⁺ is the main exchangeable cations between clay layers. Based on the results of the thermal analysis, these sediments contain kaolinite, smectite etc. We can conclude that a large amount of sodium smectite is present in these sediments. Moreover, the C.E.C values of the sediments confirm this assumption.

Based on the physical-chemical analysis on GoG sediment samples, we can conclude that the tested sediment contains large amount of exchangeable cations (Ca²⁺ and Na⁺), therefore, its cation exchange capacity is very high. In addition, the exchangeable sodium content in the sediment is relatively higher compared to that of exchangeable calcium. From the X-ray diffraction analysis, we also see that smectite is one of main mineral components in these sediments. Therefore, it is supposed that the sediment has great content of sodium smectite. In saturated conditions, compared to calcium smectite, sodium smectite has very high water uptake capacity and swelling potential. This analysis is confirmed by high water content of the sediment (Fig. 1a). Moreover, from the SEM imaging, it seems that most of aggregates show the features of card-house structure, which may be formed by arrangement of kaolinite particles with connections of smectite clay representing a plastic and bonding clay. These connections are, in fact, resulting from electrostatic attraction forces due to the presence of positive charges coming from high concentrations of exchangeable cations in the system. These positive charges are acting to balance the negative ones of very chemically and electrically (active smectite particles). This creates strong link between kaolinite aggregates. During the oedometric loading, the card-house structure degrades and then the interaggregates pores tend to be suppressed. While intra-aggregate porosities are not affected by this loading. Accordingly, it yields to decreasing dominant entrance pore sizes. Based on the interpretations above, it can be confirmed that sodium smectite plays the role of a type of “glue” or gel. The presence of smectite gel between kaolinite particles can result in the bonding effect of the sediment. It is assumed that the bonding effect described by a conceptual microstructural model of sediment (Fig. 7) can be mainly dominated by the electrostatic attraction and Van der Waals attraction forces. As result of these bonding forces, the mechanical behaviour of sediment shows the obvious cementation features during the oedometric loading.