Swelling ability and behaviour of bentonite-based materials for deep repository engineered barrier systems: Influence of physical, chemical and thermal factors

Mohammed Alzamel, Mamadou Fall, Sada Haruna

Department of Civil Engineering, University of Ottawa, Ottawa, Ontario, K1N 6N5, Canada

Keywords:Deep geological repository Engineered barrier Bentonite-sand materials Nuclear waste Swelling capacity Swelling strain

A B S T R A C T

1. Introduction

Continuous growth in the world’s population and economy,combined with rapid urbanisation,has led to an increase in global energy consumption and demand.Many countries not only face the challenge of meeting energy demands but also have to reduce the release of greenhouse gases into the atmosphere to restrict their contributions to global warming, also known as climate change.Preserving the global environment is important since the Earth is the only home that humans currently have.

Nuclear energy is a low-carbon energy resource which provides about 10% of the electricity needs worldwide, with one of the lowest carbon dioxide emissions per unit of energy produced based on the total life-cycle of energy production (International Energy Agency, 2019). However, generating electricity using nuclear power could produce a large amount of hazardous nuclear waste.Other applications of nuclear technology, such as those in the medical, industrial, and agricultural industries, also could produce radioactive waste. Since nuclear wastes would pose serious threat to human health and the environment for thousands of years,it is important to find suitable methods and techniques to safely dispose the waste. Currently, deep geological repositories (DGRs)are proposed to store radioactive waste by isolating the waste from the surrounding environments (European Commision, 2011;NWMO,2011; Yang et al., 2020).

Several countries, including Canada, believe that DGRs are the most appropriate solutions for radioactive waste management in terms of safety, economics, environmental impact, and practicality(Fall et al.,2014;Nasir et al.,2015).DGRs preserve the environment and human health by disposing radioactive waste at a depth of 300-1000 m. The waste is stored in canisters placed in a multi-barrier system that consists of natural rocks and engineered barrier systems (EBSs; e.g. waste containers, buffers, and sealing elements)(Nasir et al.,2013,2014;Yang and Fall 2021).There are many factors to be considered in the design,construction and operation of DGRs,such as the location of the repository, properties of the host rocks,design and construction of the EBS,and materials used for the waste containers(OECD,2003;Crowe et al.,2017;Guo and Fall 2020).

In Canada, DGRs based on a multi-barrier system have been studied for many years. EBSs are considered to be one of the most important components of the multi-barrier system.Bentonite-based buffer and sealing materials are critical components of the EBSs of many DGRs designs for different types of host rocks where it provides numerous important functions in a DGR system(Guo and Fall 2018,2019,2021).For instance,the barrier materials must be able to(i) contain and isolate the radionuclides, (ii) provide mechanical support to the containers,(iii)provide a thermal conductive medium to transfer heat from the waste to the surrounding natural (rocks)barrier,and(iv)seal the access shaft to the repository to ensure that it will not serve as a rapid pathway for migration of radioactive gas or groundwater to the biosphere after the closure of a repository.Moreover,the buffer materials should contribute to reducing the risk of microbiologically influenced corrosion of the used fuel containers(Zhang et al.,2019).A mixture of compacted bentonite and sand has been proposed as a potential buffer and sealing material for the EBS in Canadian DGRs as well as in other countries (Guo and Fall 2018).Fig.1 shows an example of the Canadian DGR design concept for the management of nuclear waste.

Fig.1. Conceptual design of a deep geological repository in Canada for the long-term disposal of used or spent nuclear fuel (modified from NWMO, 2012).

Bentonite is a clay that consists of mostly clay minerals from the smectite group(i.e.montmorillonite).It is considered as a suitable buffer and sealing barrier material due to its high swelling capacity,high water absorption, low hydraulic conductivity, self-sealing characteristic, and good ability to contain radionuclides if canisters fail (Cho et al.,1999; Karnland and Birgersson, 2006; Ye et al.,2010; Rawat et al., 2019). However, compacted pure bentonite has a relatively low mechanical strength.This is one of the key reasons that a bentonite-sand (B/S) mixture is a preferred material than pure bentonite to seal the access shafts.In addition to having better control on the swelling pressure, the B/S mixtures cost less than pure bentonite, and are easier to compact with higher thermal conductivity which facilitates the transfer of heat from the nuclear wastes to the surrounding rocks as opposed to pure bentonite(Villar and Lloret, 2008; Sun et al.,2009; Komine et al.,2009; Kim et al., 2011; Wang et al., 2012; Cui, 2019).

The Canadian Nuclear Waste Management Organization(NWMO) has been considering and assessing several locations in Ontario, Canada as potential DGR sites based on safety and a number of other criteria (NWMO, 2011, 2017; Crowe et al., 2017).For instance,the NWMO has been assessing whether to build a DGR at a depth of 680 m in the Municipality of Kincardine, Ontario,Canada. However, the chemical composition of the deep groundwater in the Kincardine area, like many locations in Ontario, is significantly different from that of fresh water(Jensen et al.,2009;Hobbs et al.,2011).This means that,during the lifetime of the DRG,the B/S barrier materials will react with different chemical components in the groundwater.Therefore,to ensure that the EBS will perform well in all groundwater conditions, it is important to understand whether and how significantly the swelling potential of the B/S barrier medium is affected by groundwater composition.

Numerous studies have shown that increases in the salinity of the groundwater can considerably decrease the expansion capacity of bentonite which may jeopardize its use as a barrier material(Suzuki et al., 2005; Villar 2006; Karnland et al., 2007; Castellanos et al.,2008; Zhang et al., 2012; Sun et al., 2015; Chen et al., 2015, 2017;Zheng et al.,2015,2017;Navarro et al.,2017;Liu et al.,2018;He et al.,2019;Akinwunmi et al.,2020;Shehata et al.,2021;Xu et al., 2021).For example, Chen et al.(2017) investigated the swelling behaviour of Gaomiaozi (GMZ) bentonite under different cycles of cation exchange and salinization and desalinization.The experimental results show that the salinity of the water has an influence on the swelling properties of compacted bentonite as the sodium (Na)-bentonite is partly transformed into calcium (Ca)-bentonite once the sodium chloride(NaCl)solution is replaced with a calcium dichloride(CaCl2)solution. These findings are in agreement with Chen et al. (2015)who concluded that the salinization process results in a reduction in the swelling potential of compacted GMZ bentonite. Moreover,Navarro et al. (2017) observed that there is an increase in the swelling rate of compacted bentonite with a corresponding decrease in the swelling strain when the salinity of the solution is increased.In addition, Liu et al. (2018) investigated the effects of hyperalkaline solutions on the swelling pressure of compacted GMZ bentonite by examining the sodium ion(Na+)cations and hydroxide(OH-)anions in the solution.The results indicate that high concentrations of Na+cations can inhibit crystalline swelling and double-layer swelling,and high concentration of OH-anions makes it easier for doublelayer swelling and changes in the fabric structure or arrangement of soil. Also, Akinwunmi et al. (2020) observed a decrease in the swelling pressure of Na-and Ca-montmorillonites when the salinity of the solution is increased. He et al. (2019) observed in their experimental studies that the infiltration of potassium ion(K+)salt/alkaline solutions reduces considerably the swelling pressure of compacted GMZ bentonite. However, these findings cannot be directly applied to bentonite-based materials,such as B/S mixtures,which are used in Canadian repositories as the properties of the groundwaters are different from the chemical compositions used in previous studies. Moreover, most of these studies are focused on pure bentonite.The effects of the chemical composition and salinity in the groundwater on the swelling characteristics of compacted B/S mixtures with different B/S ratios and different dry densities have not been examined in detail. Therefore, our understanding on the swelling potential of compacted B/S materials with different mixture ratios and dry densities with chemical compounds in the groundwaters in potential Ontario DGR sites is limited. There is a need to address this knowledge gap to ensure the safety of the DGR sites.

Furthermore, during the lifetime of a DGR for storing used or spent nuclear fuel or high-level waste(HLW),the barrier materials will not only be exposed to a broad range of groundwater conditions with different chemical compounds, but also subjected to high temperatures due to heat generated by the nuclear waste(Ye et al.,2013,2014).These high temperatures may intensify chemical reactions in the groundwater and induce chemical attacks on the bentonite-based barrier materials.Therefore,to ensure that the EBS will perform properly in all groundwater and thermal conditions in the Canadian’s DRGs, it is important to understand the effects of temperature and chemical reactions on the swelling properties of B/S materials. However, there is a paucity of information on the coupled effects of groundwater chemistry and temperature on the swelling capacity and mechanical behaviour of compacted B/S materials with different mixture ratios and dry densities. The objective of this study is to investigate the individual and coupled effects of the physical characteristics of the B/S mixtures(B/S ratios and densities), chemical properties of the groundwater and temperature on the swelling behaviour of the B/S mixtures used in the EBS of a DGR to store nuclear waste materials.

2. Materials and testing methods

2.1. Solid materials

The testing program examined different B/S mixture ratios(dry mass percentage of bentonite/sand:20/80,30/70,50/50,and 70/30)of commercial MX-80 Na-bentonite (from OPTA Minerals Inc.,Amherst, NY, USA) and quartz sand (GS-20 silica sand, US Silica,Katy, USA) compacted to different dry densities (ρd= 1.4 g/cm3,1.6 g/cm3, 1.8 g/cm3, and 2 g/cm3). The MX-80 bentonite is a smectite clay that predominantly consists of Na-montmorillonite.Its chemical composition is listed in Table 1. X-ray diffraction(XRD) tests were performed with a Scintag XDS 2000 diffractometer (SCINTAG, Sunnyvale, USA) to characterize the mineralogical composition of the MX-80 bentonite powder. The XRD patterns(see Fig.2)show that the MX-80 bentonite contains predominantly montmorillonite (92%) and a small amount of other minerals,including quartz, calcite, and feldspar. The main physical and chemical characteristics of the MX-80 bentonite are listed in Table 2. Sieve analysis and sedimentation tests (i.e. hydrometer tests) were carried out to determine the grain size distribution of the bentonite by following American Society for Testing and Materials (ASTM) standard D422-63, shown in Fig. 3. The analysis reveals that all the bentonite particles (fully dispersed material) can pass through the No.100 sieve(0.15 mm)and 98%pass through the No. 200 sieve (0.075 mm) (see Fig. 3). The quartz sand contains 99.99%silica(SiO2)with a specific gravity of 2.65.Based on a sieve analysis following ASTM standard D6913-04, the particle sizes of the sand range from 0.2 mm to 1.9 mm.

Table 1 Chemical composition of MX-80 bentonite.

2.2. Solutions

To investigate the effects of water chemistry on the B/S mixtures, three different water solutions were used to simulate the chemistry of different groundwater conditions from potential DGR sites in Ontario,Canada.The control solution is the distilled water,which simulates fresh groundwater close to the ground surface.The other two solutions were made to simulate the chemical compositions of the groundwater in the Trenton and Guelph regions.The main chemical characteristics (i.e. main cations and anions, pH,total dissolved solids (TDS)) of the synthetic groundwaters are listed in Table 3. Distilled water is labeled as DW, and groundwater from the Trenton and Guelph regions are referred to as the T and G solutions,respectively.CaCl2,NaCl2,and MgCl2were the main salts used to prepare the T and G solutions.

Table 2 Physical and chemical properties of MX-80 bentonite powder.

Table 3 Main chemical composition and properties of the synthetic groundwaters.

2.3. Specimen preparation and mix ratios

The MX-80 bentonite and sand were oven-dried separately at 105°C for 24 h to eliminate any moisture accumulated during storage. Then, the required quantities of bentonite and sand were thoroughly mixed with a food mixer. Four uniform dried B/S mixtures were created with different mix ratios (bentonite/sand ratios of 20/80, 30/70, 50/50, and 70/30 percent by dry mass). The grain size distribution of each mixture is shown in Fig. 3. Next,compaction tests were carried out on each B/S mixture to determine the compaction curves,optimum water contents,and maximum dry unit weights. Standard protocol compaction tests were carried out with an automatic device in accordance with ASTM D698. The soil was placed in a mold with an inner diameter of 101.6 mm and a height of 116.4 mm.The soil was compacted in three layers with the use of a 2.5 kg(5.5 lb)hammer that falls from a height of 300 mm 25 times for each layer. The required quantities of bentonite and sand were blended dry to ensure homogeneity, as suggested by Gleason et al. (1997). Then, DW was added to the B/S mixtures to reach the desired moisture content(which ranged from 5%to 25%). The samples were stored in sealed polyethylene plastic bags and kept at room temperature (23°C) for 72 h prior to compaction. The compaction curves,maximum dry densities and optimum water contents of the compacted B/S mixtures were determined.

Fig. 2. XRD pattern of the MX-80 bentonite.

Fig.3. Grain size distribution curves of the MX-80 bentonite powder,quartz sand,and bentonite-sand mixtures with different B/S mixture ratios(20/80,30/70,50/50,and 70/30).

2.4. Testing and analysis of specimens

2.4.1. One-dimensional free swelling test

The swelling induced volume changes of the compacted bentonite are illustrated in Fig. 4.

Free swelling tests were carried out on the specimens (with different B/S mixture ratios)compacted to the desired dry densities(ρd= 1.4 g/cm3, 1.6 g/cm3, 1.8 g/cm3, and 2 g/cm3). The water content and initial dry density were based on the different B/S mixture ratios.In accordance with ASTM standard D-4546-08,all of the specimens were compacted to a diameter of 63 mm and a height of 20 mm inside a polycarbonate tube (Fig. 5), which has good chemical resistance. The tube has the same diameter as the specimens with a height of 120 mm,which allows the mixtures to swell vertically but not laterally. No gap at the interface between the tube and the B/S mixtures was observed. The specimens were then soaked with 150 mL of the appropriate water solutions(DW,T,and G)under approximately zero vertical stress.A digital dial gauge was used to monitor the vertical free swelling strain of each specimen over elapsed time until the specimens achieved full saturation and ceased expanding.At the end of the swelling tests,the degrees of saturation were determined (according to the soil mechanics approach) to ensure that the specimens were fully saturated.

Fig. 4. One-dimensional free swelling mechanism (modified from Liu, 2013).

Fig. 5. Image of one-dimensional free swelling test apparatus.

The free swelling tests were conducted in a temperaturecontrolled room under two different temperatures:

(1) Some tests were carried out at a temperature of 23°C. This serves as a control temperature to simulate the thermal conditions of a DGR in which no significant heating is generated due to the nuclear waste(e.g.disposal of low level waste(LLW)).

(2) Some tests were carried out at a temperature of 80°C to simulate the heat generated by used or spent fuels or HLWs in a DGR. To avoid significant water evaporation at high temperature (80°C), the top of the polycarbonate tube was covered with a plastic lid.The plastic cover was perforated in the middle to allow the placement and displacement of the LVDT (linear variable differential transformer) without any contact with the plastic lid. Rutqvist et al. (2014) calculated that the heat released by HLWs will increase the temperature at the interface between the host rock and the bentonite to about 80°C.Moreover,they found that the temperature can be higher than 80°C in areas close to the canister. These results are also in agreement with the TOUGH-FLAC simulation results (Rutqvist et al., 2014). Furthermore, Gobien et al. (2018) also found that the temperature in highly compacted bentonite around the containers could increase to 85°C after closure of the DGR.

2.4.2. X-ray diffraction analysis

XRD analyses were conducted to determine the changes in the mineralogy of pure bentonite and the B/S mixtures after they were immersed in different water solutions and/or exposed to different temperature conditions (23°C and 80°C). The X-ray powder diffraction data were collected by using a Philips X’Pert diffractometer (Malvern, UK) with a Cu tube at an operating power of 45 kV and a current of 40 mA.The XRD patterns were determined at room temperature(23°C),and the data set was collected from 2°to 70°(2θ). The step size was 0.02°, and the counting time for each step was 0.6 s.

2.4.3. Thermal analysis

Thermal analysis was carried out on the dried pure bentonite and B/S mixture specimens after immersion in the water solutions and/or exposure to different temperatures. A TGA Q5000 V3.15 thermal gravimetric analyzer (TA Instruments, New Castle, USA)was used to perform the thermogravimetric analysis(TGA)coupled with derivative thermogravimetry (DTG) analyses on several specimens. The powder specimens (17-20 mg) were heated up to 1000°C at a rate of 10°C/min under nitrogen purge conditions(25 mL/min).

2.4.4. Mercury intrusion porosimetry tests

Mercury intrusion porosimetry (MIP) testing has been used for many decades to study the microstructure (porosity, pore size distribution(PSD))of clays.Hence,the porosity and PSD of some of the B/S specimens can be studied by using MIP. MIP tests were performed on the specimens by using a Micromeritics AutoPore III 9420 mercury porosimeter(Anton Paar,Houston,USA).AutoPore III measures the volume distribution of pores in materials by mercury intrusion or extrusion.It should be emphasized that because of the limited pressure range of the MIP technique, a significant pore volume (entrance diameter smaller than 6 nm) in the bentonite aggregates is not detectable. The MIP technique allows the exploration of pore diameter sizes between 6 nm and 600 μm. As air- or oven-drying may lead to excessive shrinkage of the clay specimen, the freeze drying method (e.g. Wang et al., 2014) was used. This method minimizes the microstructure disturbance during dehydration which improves the quality of the MIP results(Delage et al., 2006; Tang et al., 2011). In this method, a B/S specimen is cut into small pieces (approximately 1.5 g) and immersed into liquid nitrogen which has been vacuum-cooled to its freezing point (-210°C). Then, the frozen specimen is transferred to the vacuumed chamber of a freeze dryer for sublimation for approximately 24 h.

However, some researchers (e.g. Sills et al., 1973; Simms and Yanful, 2004) have questioned the applicability of the MIP test in determining the PSD of clay specimen,particularly on whether the specimen is a good representation of the materials due to its small specimen size and disturbance during specimen preparation.Another possible disturbance of the microstructure can be caused by the high injection pressure and pore accessibility. However, it should be noted that other studies(e.g.Sills et al.,1973;Lawrence,1978;Delage and Lefebvre,1984;Penumadu and Dean,2000)have concluded that the clay pore structure or microstructure is not affected by high injection pressure as the pore structure is mostly filled with incompressible mercury in the test.

2.4.5. Scanning electron microscopy analyses

The microstructure of some of the dried B/S specimens was also examined using a scanning electron microscope (SEM). SEM analysis was performed by using a JSM-6610LV scanning electron microscope (JEOL, Peabody, USA) to study the pore structure and morphology of the specimens.

3. Results and discussion

3.1. Effect of the physical factors of the bentonite-sand barrier materials in distilled water

The swelling strain versus time curves of the compacted B/S mixtures(with different mixture ratios and dry densities)exposed to DW at a temperature of 23°C are shown in Fig. 6. The final or maximum swelling strains of the B/S specimens with different initial dry densities are shown in Fig. 7. In this study, the swelling strain (ε, in percent) is calculated using

Fig. 6. Effect of dry densities on the swelling strain of (a) 70/30 bentonite-sand mixtures and (b) 30/70 bentonite-sand mixtures immersed in DW.

Fig. 7. Final swelling strain for bentonite-sand specimens with different mix ratios at different initial dry densities.

where ΔHis the swelling deformation or change in height andH0is the initial height of the specimen.

As shown in Figs.6 and 7,it can be seen that the swelling rate and the maximum swelling strain depend on both the initial dry density and the mix ratio. Fig. 6 shows that, for all cases in this study, the maximum swelling strains are reached between 48 d and 88 d.Specimens with lower initial dry densities or higher sand contents reach the maximum swelling strain earlier. It is also observed that,regardless of the initial dry density and blending ratio,the measured swelling strain follows a sigmoid relationship with logarithm time(Fig.6).The initial increase in strain is slow and less pronounced for the specimen with a lower mixture ratio (B/S: 30/70). This is followed by a rapid increase in strain with time which represents approximately 80%of the total observed deformation.Subsequently,the strain increases slowly until reaching an asymptotic value (the maximum swelling strain). From Fig. 7, it is obvious that the maximum swelling strain increases as the initial dry density or the mixture ratio increases. For a 70/30 (B/S) mixture ratio, the maximum swelling strains are 266%, 277%, 286%, and 304% for dry densities of 1.4 g/cm3,1.6 g/cm3,1.8 g/cm3,and 2 g/cm3,respectively.For a 30/70 (B/S) mixture ratio, the maximum swelling strains are 137%,163%,176%,and 191%for dry densities of 1.4 g/cm3,1.6 g/cm3,1.8 g/cm3,and 2 g/cm3,respectively.Fig.7 shows that the maximum swelling strain and the dry densities follow a linear relationship,εmax=aρd+b(R2>0.97),whereaandbare the fitting parameters,with a positive slope over the ranges of densities.This confirms that increase in the dry density of the B/S increases the swelling potential.

The observed sigmoidal behavior of the swelling strain versus logarithm time is the result of the swelling process of the mixtures of bentonite and coarser non-swelling particles (sand particles in this study).Indeed,it is well known that the swelling mechanism of B/S mixtures can be divided into three stages: initial swelling (inter-void swelling), primary swelling, and secondary swelling(Suzuki et al.,2005;Rao and Thyagaraj,2007;Cui et al.,2012).The initial swelling occurs when a B/S specimen comes into contact with water and expansive minerals such as montmorillonite swell into voids (i.e. macropores) created by the coarser, non-welling sand particles. The swelling of montmorillonite into the macropores will depend on the boundary conditions as well as the dry density of the material. However, at a low bentonite content (i.e.low mixture ratio in this study), the inter-particle voids of the skeleton occupy large space,in which the bentonite can swell freely during the absorption of water without inducing a significant amount of expansion during the initial phase of swelling(Sun et al.,2013). This means that the swelling potential of the bentonite is reduced by filling the inter-particle voids. Therefore, the swelling strain of the specimens with low B/S ratio is less pronounced (see Fig. 6b). Most of the swelling occurs during the primary stage of swelling which is caused by the hydration of the cations exchanged between the crystal layers. A sudden change in the swelling curve at 0.1 d in Fig. 6b marks the beginning of the primary stage of swelling. The secondary stage of swelling occurs when crystalline swelling is completed and a double layer is created. During this stage, swelling follows a linear relationship with logarithm time(Suzuki et al., 2005; Cui et al., 2012,2019; Rawat et al.,2019).

The dependency of both the maximum swelling strain and swelling rate on the mixture ratio and initial dry density can be explained by the fact that the swelling potential of the B/S mixture is highly related to the volume of bentonite in the unit volume of mixture(Komine et al.,2009;Cui et al.,2012;Shehata et al.,2021).This means that, for a given dry density, an increase in the bentonite content or B/S mixture ratio will lead to an increase in the volume of bentonite per unit volume of the mixture, therefore increasing the swelling strain and the swelling rate. This explanation is consistent with the results from the SEM analyses presented in Fig.8.This figure shows the SEM images magnified 14 times for B/S mixtures of 30% (B/S: 70/30) and 70% (B/S: 30/70) sand contents.It is seen that,with an increase in the bentonite content,the sand particles are further apart.In other words,there is an increase in the volume of bentonite per unit volume of the mixture materials.

Moreover, for a given B/S mixture ratio, a higher dry density is associated with a smaller volume of voids(smaller size of pores)in the B/S mixtures. Consequently, the volume ratio of bentonite to void in the mixtures increases with increasing dry density of the B/S mixture materials,which in turn increases the swelling strain and swelling rate.The refinement of the pore structure or reduction of the volume of voids in the B/S mixtures due to higher dry density is experimentally supported by the results from the MIP tests performed on two specimens with a mixture ratio of 70/30 (B/S) and dry densities of 1.63 g/cm3and 2 g/cm3(see Fig. 9), respectively.This figure shows that an increase in the dry density does not change the range pore sizes but it decreases significantly the amount of intruded pores (Fig. 9). Moreover, the specimen with a higher dry density has a lower total porosity(Fig.9b)and finer PSD(Fig. 9a) than that with a lower dry density.

Fig. 9. Results of MIP tests on two specimens with 70/30 bentonite-sand and dry densities of 1.63 g/cm3 and 2 g/cm3: (a) Incremental porosity, and (b) Cumulative porosity.

3.2. Influence of the chemistry of groundwaters on the bentonitesand mixtures with different dry densities and mixture ratios

The changes in swelling strain with time for the compacted B/S mixtures saturated with DW, G and T solutions with different concentrations of TDS are shown in Figs. 10-13. Figs. 10 and 11 show the relationship between the swelling strain and time for the B/S mixtures with the same initial dry density(1.63 g/cm3)and different mixture ratios in T solution and DW (Fig. 10) and G solution and DW (Fig. 11), respectively. Figs. 12 and 13 show the effect of the chemistry of the T and G solutions on the swelling strain with different mixture ratios and initial dry densities.

Fig. 10. Effect of distilled water and T solution on the swelling strain of bentonitesand mixtures with different mixture ratios and a constant initial dry density(1.63 g/cm3): (a) DW, (b) T solution, and (c) Comparison between DW and T solution.

Fig.11. Effect of distilled water and chemistry of G solution on the swelling strain of bentonite-sand mixtures with different mixture ratios and a constant initial dry density (1.63 g/cm3): (a) DW, (b) G solution, and (c) Comparison between DW and G solution.

Fig.12. Effect of chemistry of T solution and initial dry density on the swelling strain of: (a) 70/30 bentonite-sand mixtures, and (b) 30/70 bentonite-sand mixtures.

The results show that, in general, at low dry densities, the swelling rate is slow but increases gradually with increase in dry density, regardless of the chemistry of the solution. Moreover,swelling develops more rapidly in specimens with higher dry densities, irrespective of the type of water. Furthermore, the swelling rate is time-dependent. Initially, the increase in swelling deformation with time is slow.After that,swelling increases rapidly with time followed by a decrease in the strain rate until reaching the maximum swelling strain. The mechanisms responsible for these behaviours have been discussed earlier.

The results here indicate that the chemistry of the studied Ontario’s groundwaters has a substantial effect on the swelling strain,swelling rate and maximum swelling strain(Fig.14)of the B/S mixtures,regardless of the initial dry density and bentonite content.However, the magnitude of this effect is a function of the chemical composition of the groundwater (particularly the TDS), bentonite content (the B/S mixture ratio) and the initial dry density (Fig.15).The specimens in DW exhibit the highest swelling rate and require the longest period of time to reach the final stage of swelling,while the specimens in T and G solutions have a lower swelling rate and shorter total swelling time (Figs. 10-15). The maximum swelling strain is reached at 20-38 d for the specimens in the G and T solutions,and 48-88 d for the specimens in DW,depending on the dry density, TDS, and mixture ratio. Also, the mixtures saturated with DW have a greater maximum swelling deformation compared with the mixtures saturated with T and G solutions (Figs. 10-15). This means that when the volume of total dissolved solids is increased,the initial swelling rate, swelling rate during the primary stage of swelling and the maximum swelling strain are decreased (Fig.15).Moreover,the swelling strain is decreased significantly with increase in TDS concentration.

This chemically (salinity) induced deterioration of the swelling ability of the B/S mixture specimens is consistent with the results of the MIP tests(Fig.16).MIP tests were performed on two specimens with a mixture ratio of 70/30(B/S)and dry density of 1.63 g/cm3in DW and G solutions after 40 d. In Fig.16,the specimen with the G solution shows higher incremental intruded pore volume and coarser pore structure than the specimen with DW solution.However, most of changes in the microstructure take place in the category of macropores. For the specimen with larger swelling strain or swelling capacity, i.e. the specimen with DW solution,more interlayer hydration has occurred, resulting in a constriction of the accessible pores. These observations on the chemically induced changes of the pore structure of the MX-80 B/S mixtures are in agreement with the findings in Mata(2003)who concluded that saline water has a significant influence on the macropores of bentonite-based mixture materials,whereas its effect on the micropores is not significant.

The observed deterioration of the swelling characteristics(e.g.swelling strain and swelling rate,maximum swelling strain)of the B/S mixtures saturated with T and G solutions,irrespective of the dry density and B/S mixture ratio,is related to the high TDS values of the T and G solutions. Higher dissolved salt concentrations or TDS reduce the swelling ability of bentonite (Suzuki et al., 2005;Karnland et al.,2007;Castellanos et al.,2008;Komine et al.,2009;Zhang et al., 2012; Shehata et al., 2021). This reduction in the swelling potential of the bentonite saturated with the G or T solutions can be explained by the two following mechanisms.First,a higher volume of TDS causes a reduction in the thickness of the double diffused layer (DDL) of the montmorillonite particles which leads to a decrease in the repulsive forces between the clay particles. Consequently, this reduces the swelling strain and swelling pressure of the materials(Siddiqua et al.,2011).Secondly,chemical changes take place in the montmorillonite minerals as the chemical attacks from the saline G and T solutions. In other words, the Na-montmorillonite minerals transform, at least partially, to Ca-montmorillonite minerals during interaction with Ca2+ions in the G or T solution, as evidenced from the results of the microstructural analyses (XRD, TG/DTG (thermogravimetry/differential thermogravimetry)) shown in Fig. 17 (XRD patterns)and Fig. 18 (TG/DTG analysis results). XRD and TG/DTG analyses were performed on 30/70 bentonite-sand mixtures saturated with DW and G solution, respectively. The results of the microstructural analyses indicate that cation exchange took place in the specimens saturated with the G solution (G23)in the presence of divalent cations (Ca2+, Mg2+) in the porewater of the bentonite(Table 3), which in turn led to the transformation of Na- to Camontmorillonites.The Ca2+cation,has a much greater affinity for montmorillonite than the Na+cation, and the substitution of the Na+cation in the Na-montmorillonite creates the Camontmorillonite with reduced swelling capacity (Akinwunmi et al., 2020; Shehata et al., 2021). From Fig.17, it is seen that the bentonite-based materials consist predominantly of montmorillonite, with small proportions of illite and quartz. Moreover, the spacing between the crystal interlayers of bentonite (i.e. basal spacing) is sensitive to the hydration state of the ions, and the position of the peak in the basal spacing (d???) is affected by the type of interlayer cation.Therefore,the XRD spectrums of the DW specimen reveal that the first peak of the Na-montmorillonite is at 8.92°(2θ) andd???= 9.9 ?, while the G specimen (G23) is at (2θ)6.28°with lower reflection intensity andd???= 14.04 ?. These changes in the position of the peak and the reflection intensity indicate a reduction in the swelling capacity(Moore and Reynolds,1989). These XRD results are in agreement with those of the TG/DTG analysis in Fig.18. From Fig.18a, it can be seen that the TG/DTG results of the specimen mixed and saturated with DW indicate that the initial loss of the mass of water of 2.64 wt%occurs at a low temperature interval(0-50°C)because of the removal of the surface water. The second loss of the mass of water (1.8 wt%)occurs within the temperature interval of 200-600°C, which corresponds to water desorption from the interlayers. The last endothermic peak, within the temperature interval of 600-1000°C, has a mass loss of about 3.22 wt%, due to the dehydroxylation of the O-H structure and collapse of the smectite crystal structure (Drits et al., 1998; Frost and Ding, 2003;Boudriche et al., 2012). Fig.18b clearly shows that the total mass loss of the specimen mixed and saturated with G solution within a temperature interval of 50-200°C is 18.2 wt%. In addition, the interlayer of chemisorbed water is gradually removed at a temperature interval of 200-600°C, therefore producing a mass loss of 3.85 wt%.Dehydroxylation of the Al-OH bond structure occurs at a higher temperature interval (600-1000°C) with an endothermic peak of 9.8 wt%.Therefore, these TG/DTG results confirm that the B/S microstructure is partially altered in a highly saline solution.

Fig.13. Effect of G solution and initial dry density on the swelling strain of: (a) 70/30 bentonite-sand mixtures, and (b) 30/70 bentonite-sand mixtures.

Fig.14. Effect of different solutions on the dry density and swelling strain.

Fig. 15. Comparison of the swelling strain of different bentonite-sand ratios with a density of 1.63 g/cm3 and exposed to G and T solutions.

However,Figs.12-14 show that the reduction in swelling of the saline solutions is smaller for bentonite-based materials with higher dry densities.This reduction in swelling with increase in dry density can be explained by the mechanisms of crystalline swelling and double-layer swelling of the bentonite. Crystalline swelling results from the hydration of exchangeable cations, e.g. K+, Na+,Ca2+,and Mg2+,between mineral layers that show a structure with one alumina octahedral sheet sandwiched between two silica tetrahedral sheets. This is a mechanism that the adsorption of the maximum volume of hydrates depends on the nature of the cations(Wang et al., 2014). When three to four water monolayers are formed, the surface hydration mechanism plays a less important role and electric double-layer repulsion becomes the key swelling mechanism (Bradbury and Baeyens, 2003; Suzuki et al., 2005;Wang et al., 2014). For bentonite with high dry density, the low quantity of water adsorbed is essentially pseudo-crystalline interlayer water.Hence,the latter is not enough to build the DDL(Pusch,2000; Wang et al., 2014). Consequently, swelling is largely controlled by the crystalline swelling and the double-layer repulsion provides only a slight contribution to the swelling.In this case,bentonite-water interaction is essentially governed by the exchangeable cations (Suzuki et al., 2005; Castellanos et al., 2008;Siddiqua et al., 2011).

Fig.16. Pore size curves: (a) MIP cumulative porosity curves, and (b) MIP PSD (70/30 bentonite-sand;dry density of 1.63 g/cm3;one specimen saturated with DW solution,one specimen saturated with G solution; soaking time of 40 d).

3.3. Coupled effect of the chemistry of groundwaters and temperature on the bentonite-sand barrier

To properly design the experimental testing program and address the objective of this section as well as better analyze the results, preliminary investigations were conducted to assess the individual impact of temperature on the swelling capability of the bentonite-sand barrier materials saturated with DW (i.e. no chemical effect considered). One-dimensional free swelling tests were conducted on compacted B/S material specimens with different mixture ratios in DW and exposed to different temperatures (23°C and 80°C). The results indicate that temperature has insignificant effects on the swelling potential of the materials.Indeed, the maximum amount of swelling of the specimens with DW at 23°C is only 2% higher than that of the 80°C specimens.These results agree well with the results from previous studies(e.g.Pusch, 2000; Shehata et al., 2021) that show insignificant changes in the swelling potential of bentonite based materials with increase in temperature(up to 90°C).However,the following questions still remain.Does temperature have a significant effect on the swelling ability of bentonite-based materials saturated with G solution? In other words,how significant is the coupled effect of the chemistry of groundwaters and temperature on the swelling properties of bentonite-sand barrier materials?To address these questions,free swelling tests were carried out on compacted specimens with different mixture ratios(B/S:70/30,50/50)and saturated with the G solution at room and higher temperatures of 23°C and 80°C,respectively. Typical results are shown in Fig. 19. It is seen that,regardless of the mixture ratio, the swelling strain is significantly dependent on the temperature and chemical composition of water.Higher temperature results in more intense chemically induceddegradation of the swelling capacity of the materials. It is obvious that the specimens with the G solution at 80°C have a swelling strain that is lower than those at 23°C with shorter time to reach the stage of constant swelling rate and termination of swelling.For example, Fig. 19 shows that the specimen (B/S: 70/30) tested at 80°C gives a swelling strain of 17% in 25 d, while the specimen exposed to a temperature of 23°C shows a swelling strain of 30%in 30 d. In other words, the swelling potential of the (B/S: 70/30)specimen in G solution at 80°C is decreased by 43%in comparison with that of the specimen at 23°C.Also,the same trend is observed in the specimen with a mixture ratio of 50/50 (B/S) which gives a swelling strain of 7%in 26 d at 80°C,while the specimen at 23°C exhibits a swelling strain of 13%in 30 d.The test results show that higher temperatures intensify the chemical reaction of the bentonite in G solution which reduces the swelling potential of the B/S mixtures. Future swelling stress experiments should be performed to determine how this significant loss in free swelling strain translates into loss in swelling stress, and what may be the implications for repository performances. This temperature-induced intensification could be related to the following mechanisms: (i)higher temperature increases the dissolution rate of the smectite,which would reduce the amount of expansive minerals (montmorillonites)in the B/S mixtures;(ii)higher temperature increases the transformation of Na-montmorillonite minerals to Camontmorillonite which has a lower swelling potential; and (iii)higher temperature leads to lattice contraction in different clay structures.

Fig.19. Coupled effect of the chemistry in G solution and temperature on the swelling of 70/30 bentonite-sand mixtures and 50/50 bentonite-sand mixtures(dry density of 1.6 g/cm3).

Indeed,a number of previous experimental studies on smectite dissolution kinetics or rates of dissolution in different chemical(pH,different chemical solutions) and thermal (temperature) environments have reported that the dissolution rate of smectite increases with temperature. This matches the findings from previous numerical studies (Zheng et al., 2015, 2017; Xu et al., 2021). For example,Amram and Ganor(2005)investigated the coupled effect of pH and temperature (25-70°C) on the dissolution rate of smectite SAz-1 by conducting dissolution experiments with the use of non-stirred flow-through reactors in thermostatic water. Their results show that the smectite dissolution rate is increased with temperature and decreased with pH value.Moreover,Huertas et al.(2001) evaluated bentonite dissolution in granitic solutions (pH value of 7.6-8.5)in a semi-batch reactor at 20°C,40°C and 60°C.They concluded that temperature increases the dissolution rate of smectite and the effect of temperature on the dissolution rate is a function of the activation energy.These findings on the reduction of the amount of montmorillonites due to a higher dissolution rate at a higher temperature are in agreement with the results of the XRD analyses presented in Fig.17.Fig.17 shows the XRD patterns of the B/S mixture in G solution at room temperature (23°C, G23) and higher temperature of 80°C(G80)as well as the XRD pattern of the DW sample. It can be observed that the first peaks for the montmorillonite in the DW,G23 and G80 samples are(2θ)8.92°,6.288°and 6.632°,with lower reflection intensities and the basal spacings ared001= 9.902 ?, 14.044 ?, and 13.1357 ?, respectively. This detected shift to the left of the first peak of the G80 specimen is related to the transformation of Na- to Ca-montmorillonite,

Fig.17. XRD patterns of the 30/70 bentonite-sand mixtures mixed and saturated with(a)DW at 23 °C(DW23),(b)G solution at 23 °C(G23),and(c)G solution at 80 °C(G80).

Fig. 18. Results of TG/DTG analyses of a 30/70 bentonite-sand mixtures mixed and saturated with (a) DW, and (b) G solutions.

whereas the observed lower intensity indicates a decrease in the amount of minerals that undergo swelling. The latter suggests a higher dissolution of montmorillonite in the G80 specimen. An additional mechanism can be suggested as a factor that has contributed to the higher decrease in swelling ability observed in the specimens saturated with G or T solution at high temperatures(80°C). The phenomenon can be explained in terms of lattice contraction at high temperature (Ureana et al., 2013; Chen et al.,2018). Chen et al. (2018) carried out swelling and microstructural tests to investigate the combined effects of temperature(20-60°C)and cation type (in the NaCl and CaCl2solutions) on the swelling pressure and microstructure of densely compacted pure GMZ bentonite.They found that lattice contraction at high temperatures decreases the swelling pressure to a greater degree in CaCl2-bentonite with smaller basal spacing. Due to the heating of the saturated CaCl2-bentonite, there is a change of the interlamellar adsorbed water to bulk water which weakens crystalline swelling.The XRD results obtained in this study show that the basal spacingd001of the G80 specimen(13.1357 ?)is smaller than that of the G23 specimen(14.044 ?).In other words,there is a decrease in the basal spacing with temperature. However, it should be noted that more detailed microstructural and mineralogical investigations should be carried out in the future to gain more insight into the changes that take place in the microstructure and mineralogy of compacted MX-80 B/S due to the combined effects of temperature and groundwater chemistry.

4. Conclusions

Based on the results in this study,the following conclusions are made:

(1) The swelling potential of the studied compacted B/S mixture is significantly dependent on both its initial dry density and bentonite content. Higher dry density and/or larger B/S mixture ratio leads to higher swelling strain of the materials.The swelling strain follows a sigmoid relationship with logarithmic time, irrespective of the initial dry density and B/S mixture ratio. The maximum swelling strain and the initial dry density follow a linear relationship for densities between 1.4 g/cm3and 2 g/cm3.

(2) The chemical composition of the studied groundwaters decreases significantly the swelling potential of the B/S barrier materials,irrespective of its initial dry density and bentonite content.However,the magnitude of this decrease in swelling depends on the volume of TDS in the solution. The B/S mixtures with more sand are better able to reduce the effects of chemicals in the water on the amount of swelling.

(3) The chemically-induced reduction of the swelling potential of the B/S barrier materials is caused by(a)the reduction of the DDL thickness of the montmorillonite particles due to a high volume of TDS and(b)the transformation of Na-to Camontmorillonites with less swelling potential due to cation exchange.

(4) The swelling rate and swelling potential of the B/S mixtures are significantly reduced, not only due to the increase in salinity of the water (chemical factor), but also affected by the combined effects of the chemical factor and the temperature(thermal factor).Higher temperatures lead to more intense chemically induced-deterioration of the swelling performance.

(5) It is recommended that,in the conceptual design of EBSs for HLW repositories in Ontario or Canada, not only the chemistry of the deep groundwater at potential DGR sites,but also the temperature expected within the EBS should be carefully considered to meet the design criteria with respect to swelling potential and swelling pressure of the selected bentonite-based barrier materials.

The findings presented in this paper will contribute to more costeffectiveandsaferdesignofEBSandDGRsystemsinCanadaandother regions with similar groundwater chemistry.Despite these results,it is still necessary to investigate the swelling potential and swelling characteristics of compacted B/S mixtures at high temperatures(85°C,100°C and 150°C) in groundwater solutions with different chemical compounds. Indeed, Rutqvist (2020) computed temperaturesashigh as150°C atthe interface between canister and bentonite fordensecanister spacingusinglargedisposalcanisters.Furthermore,experiments are being carried out at the Mont Terri Laboratory(St-Ursanne, Switzerland) with temperatures in the bentonite buffer of up to 140°C (e.g. Garitte et al., 2017). Moreover, it is necessary to assess the coupled effect of the chemistry of deep groundwater and temperature on the swelling pressure or stress of compacted B/S mixtures.This will be the object of future studies.Finally,this study focusedonthefreeswellingexperiments.Inotherwords,theswelling stresses of the tested specimens were not evaluated.However,both parameters, percentage of free swelling and swelling stress, are important performance criteria.Investigating the swelling stress was outside of the scope of this project. Swelling stress experiments,which are much slower and more expensive than free swelling experiments,will be considered in future studies.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We acknowledge the funding support from Natural Sciences and Engineering Research Council of Canada (NSERC).