Rock salt and depositional environment

Rock salt is a rock that is mainly composed of salt, generally halite (NaCl). This term is used in mining, engineering and industry settings, whereas evaporite is commonly used in geological (sedimentologic, diagenetic) context. The major rock salt deposits consist of halite (NaCl), anhydrite (CaSO4), and gypsum (CaSO4 · 2H2O), but also other minerals such as sylvite (KCl), polyhalite (K2Ca2Mg(SO4)4 · 2H2O), carnallite (MgCl2 · KCl · 6H2O). Halite and sylvite are colorless in their pure form, but they can show a color due to the presence of trapped air bubbles, fluid inclusions, finely dispersed clay (Figure 1), organic matter, hematite, KBr, and lattice defects (Sonnenfeld, 1995). The clay present in rock salt has a detrital origin, and is commonly illite, kaolinite, chlorite, smectite or mixed-layer clay (Warren, 2006). The variation in salt mineralogy is linked to its depositional setting and diagenetic environment. Due to chemical precipitation, rock salt forms a saturated saline solution at the surface, generally in saline lakes in arid or semi-arid environments. The composition of rock salt is thus related to the depositional setting and the hydrochemical nature of the parent brines (Osichkina, 2006). The texture of the rock salt, however, is generally secondary due to diagenetic processes in shallow burial or postuplift settings, rather than primarily associated with the depositional setting, since rock salt dissolves easily (Warren, 2006).

Evaporation of seawater and other saline fluids will lead to the precipitation of certain salts in a specific order based on the composition of the parent fluid and environment. For the evaporation of current seawater, it is determined that salt precipitation proceeds in the following order: aragonite (at twice the concentration of seawater), gypsum (at 4–5 times the concentration of seawater), halite (at 10–12 times the concentration of seawater), and then potassium or magnesium sulfates and chlorides (predominantly carnallite and epsomite), when supersaline fluids reach concentrations of 70–90 times that of seawater (Carpenter, 1978; McCaffrey et al., 1987) (Figure 2). Geological rock salt layers may present a different order since the composition of seawater has changed through time (Warren, 2006). Rock salt is not only formed through evaporation of seawater, but in saline lakes with a different mineral assemblage, such as sodium carbonate minerals (trona, nahcolite, shortite), sodium silicate minerals (magadiite and kenyaite) and sodium or calcium borate minerals (Smoot & Lowenstein, 1991). Saline lacustrine deposits are commonly classified based on salinity, or they can be distinguished as carbonate or sulfate types with certain composition and distribution of salt and clay minerals (Qi et al., 2021). Besides evaporation, additional processes can lead to the formation of rock salt, namely temperature changes (Sloss, 1969), mixing of brines (Raup, 1982), and brine freezing (Sonnenfeld, 1984).

Methods of mineralogical identification

The most common method for mineralogical identification of rock samples is powder X-ray diffraction (PXRD) analysis. The principle behind the PXRD method is the Bragg's equation, which presents the relation between the angle at which constructive interference occurs ( θ) between waves of wavelength ( λ) and the distance between atoms on lattice planes in a crystal ( d):

(1)

Angles of constructive interference are measured by an electronic detector that records the intensity of the signal while the sample is scanned over a range of diffraction angles. The PXRD method is used to identify the mineralogical assemblage of samples, as it provides information on the crystal structure of the sample constituents based on the diffraction pattern that is characteristic for the position and type of atoms in crystals (Figure 3). The diffractograms are analyzed and quantified using PXRD software with reference databases of ICCP PDF series and Rietveld analysis. The PXRD method cannot be used for the detection of very small amounts or trace contents of certain minerals in samples, since those signals may not be distinguishable from the background noise or from the diffraction peaks of the major constituents.

The preparation of samples for PXRD analysis varies and depends on the amount of available material, bulk analysis versus analysis on separated fractions, and randomly oriented versus oriented mounts. Common PXRD analysis is performed on randomly oriented bulk sample, which is prepared by crushing the rock sample to powder, and placing it in a holder for PXRD measurement. Because of the small size of clay particles, the signal from the clay in samples can be fairly weak in nonoriented bulk samples. Therefore, additional sample preparation is carried out to separate the clay fraction from the bulk sample, using ultrasonication of the sample in an aqueous solution with sodium hexametaphosphate to disperse the particles, and centrifugation to separate the heavier particles of the sample from the supernatant with clay. Oriented clay mounts are then prepared by gravitational settling of the clay from the clay suspension on the mounts. Analysis of the clay mounts by PXRD provide clear reflections, and further differentiation between clay minerals is facilitated by additional chemical treatments in terms of swelling of clays or dissolution in hydrochloric acid or by heat treatments (Moore & Reynolds, 1997).

Another conventional method of mineralogical identification of rock samples is petrographic analysis using an optical polarizing light microscope. Such analysis is conducted on 30 µm thin slices of rock by passing light through the sample. Different minerals interact differently with light due to their different internal structure and composition, and this enables identification of the minerals. Using this petrographic method with magnification, not only major constituents but also minor or trace minerals in the sample can be identified, as well as the rock texture.

Automated mineral identification systems linked to scanning electron microscopes include the Mineral Liberation Analyser (Ferrari et al., 2021; Ji et al., 2021) and QEMSCAN instruments (Ma et al., 2016; Zhao, Zhang, et al., 2022). The method is based on high-resolution backscatter electron image analysis, energy dispersive X-ray analysis and automated microscope operation and data acquisition, and is usually employed on thin sections or polished thick sections. The technique involves automated collection of mineralogical maps of samples. Moreover, it can generate large, statistically representative datasets of porosity, grain size and shape, mineralogical associations and digital textural maps. A disadvantage of the technique is that polymorphs, such as calcite, aragonite and vaterite, cannot be distinguished, and it is also difficult to distinguish between minerals with very similar chemical compositions, such as gypsum and anhydrite. Differentiation of polymorphs and minerals with similar composition can be easily made using PXRD, as discussed above.

Methods of geochemical characterization

Geochemical characterization of rock salt can be related to the mineralogical composition, and it provides additional information on impurities within phases, and data which can help identify the origin of the brine that precipitated the salt.

Powdered rock samples can be dissolved using aqueous (acidic) solutions, and then the concentrations of the cations in the solutions are measured by inductively coupled plasma optical emission spectrometry (ICP-OES) or mass spectrometry (ICP-MS). An inductively coupled plasma is generated by ionization of argon gas in an intense electromagnetic field, resulting in excited atoms and ions that emit radiation of characteristic wavelengths. The intensity of the emissions at specific wavelengths is related to the concentration of specific elements, and these can be quantified by using a calibration series. With OES, concentrations can be measured down to parts per million, whereas MS can be employed to measure concentrations down to parts per trillion. Measurements of anions in the fluids are conducted using ion chromatography. With anion-exchange chromatography, anions are separated based on their affinity to bind with a positively charged stationary ion exchanger by Coulombic interactions. The concentration of anions in fluid samples is determined through comparison of signals with calibration fluids of a series of concentrations.

Geochemical elemental analysis can also be conducted on the (powdered) rock sample, without the need for dissolution, by X-ray fluorescence (XRF) spectrometry (Hancerliogullari & Eyuboglu, 2020; Quye-Sawyer et al., 2015). In this case, the sample can either be a polished surface of a rock sample, powdered rock sample, or a pressed pellet made from powdered rock. XRF is a nondestructive technique, whereby a sample is irradiated by high energy X-rays that cause excitation and ejection of electrons from an inner orbital shell of atoms in the sample. As a consequence, an electron from a higher orbital shell will fall down to fill the created vacancy in the lower shell, and with this, a fluorescent X-ray is emitted. The emitted radiation is characteristic for the transition between shells of specific elements, which enables identification of elements in the sample, and the intensities of the radiation allow quantification of the concentrations of these elements.

Geochemical elemental analysis of polished sections or thin sections can be performed by electron probe microanalyzer (EPMA). EPMA works in a similar way to a scanning electron microscope. The sample is bombarded with an electron beam, which results in the emission of X-rays at wavelengths that are characteristic for specific chemical elements. This method allows for high resolution analysis and automated geochemical mapping of samples. In contrast to the above methods, EPMA enables very detailed small-scale analysis rather than bulk analysis of samples (Vandeginste et al., 2020). For example, geochemical zoning of minerals can be documented and mapped at micrometer scale by EPMA.

Geochemical isotopic methods such as sulfur, oxygen, chlorine and bromine isotopes have been applied to evaporites (Ding et al., 2019). These isotopic analyses are used to identify the source of the brine that has formed the rock salt and to distinguish between marine and nonmarine origin. Fluid inclusions in salt crystals can provide more information on the formation temperature and geochemistry of the brines (Wang et al., 2016), similar to how information is derived from fluid inclusions in other minerals (Vandeginste et al., 2017).

Besides the general geochemical characterization of rock salt, it is also important to determine the water content, in particular in studies that involve geomechanical rock salt analyses. Water is present in rock salt samples as interstitial brine, in hydrated minerals and as fluid inclusions within crystals (Jockwer, 1982; Roedder & Bassett, 1981). In situ, resistivity analysis can be used to predict water content in rock salt. In small rock salt samples, thermogravimetric analysis (TGA) is employed to determine and quantify the types of water in the samples (De Las Cuevas & Pueyo, 1995). The TGA measurement records the changes in the mass of the sample while the temperature increases over time. In addition, Fourier transform infrared (FTIR) spectrometry has been used to measure water content in rock salt samples, and a calibration for this was established with synthetic samples (Ter Heege et al., 2005). Using this technique, high spectral resolution data are generated for the infrared spectrum of adsorption or emission of a sample. The wavenumber of 1650 cm−1 corresponds to the bending vibration of the water molecule in a saturated NaCl solution.

Deformation microstructures and mechanisms

Three deformation mechanisms and corresponding microstructures (Figure 4) can be distinguished: (1) cataclastic deformation with the development of microfractures and rotation of displacement of grains without permanent lattice distortion, (2) diffusive mass transfer with removal, transport and deposition of material with permanent lattice distortion or melting, and (3) intracrystalline plastic deformation with permanent lattice distortion without fracturing (Blenkinsop, 2002). One of the characteristic features of rock salt is its plastic deformation behavior that occurs already at relatively low temperature and low confining pressure in comparison to that for other types of rocks (Urai & Spiers, 2007).

First, the cataclastic deformation mechanism involves microcracking or frictional sliding. Microcracks are called intragranular when they occur within single grains, transgranular when they cut through multiple grains, or circumgranular when they occur at grain boundaries (Blenkinsop, 2002). Brittle deformation with microcrack development, grain rotation and dilational behavior occurs in rock salt only at very low effective confining pressure and high deviatoric stresses (Cristescu & Hunsche, 1998; Peach & Spiers, 1996; Peach et al., 2001). This dilation domain is separated from the compaction domain with ductile, creep deformation by the dilatancy boundary (Cristescu & Hunsche, 1998; Van Sambeek et al., 1993). Microfracture development in the dilation domain results in an increase in permeability in rock salt (Van Sambeek et al., 1993). Deformation induced by microcracking at low effective mean stress can thus lead to the formation of percolating pore space, commonly observed in the zone of disturbed rock around openings in salt mines of nuclear waste repositories, and in high overpressure underground environments (Ghanbarzadeh et al., 2015). Cyclic loading tests on rock salt have indicated that the number of developed intergranular cracks increases with the ratio of maximum cycling stress to the uniaxial compressive strength (Zhang, Wang, et al., 2022). Moreover, the tests show that the number of large pores and total pores increase and the number of small pores decrease after 12 000 cycles (Zhang, Wang, et al., 2022).

Second, diffusive mass transfer is a deformation mechanism that, at temperatures of up to about 200°C in the diagenetic domain, proceeds via solution and precipitation, resulting in textures with corroded grain surfaces and overgrowths. The mechanism involves material removal in the direction of maximum pressure leading to fabrics with indented, truncated and interpenetrating grain boundaries, and material addition in the direction of minimum pressure resulting in textures with overgrowths (Blenkinsop, 2002). Thus, grains dissolve at highly stressed contact points and form a brine fluid film, and this material diffuses towards the zones of lower stress, where the material then precipitates (Schutjens & Spiers, 1999). This dissolution, diffusion transport, and precipitation is driven by the difference in chemical potential between the highly stressed contact grain boundaries and the intergranular zone under lower stress. Diffusive mass transfer can also result in fluid inclusion planes that are perpendicular to the direction of minimal stress. Moreover, studies have shown that diffusion can occur at much lower temperatures and lower differential stress in wet rock salt than in dry rock salt (Urai et al., 1986). Diffusive mass transfer microstructures are distinguished from cataclastic microstructures due to the lack of microfractures. And they are different from intracrystalline plastic deformation microstructures, for they do not have undulatory extinction, subgrains or recrystallized grains in their fabrics. As previous studies have shown, the diffusive mass transfer process can be linked to the intergranular plastic deformation process (Urai & Spiers, 2007). Pressure solution creep is driven by diffusive mass transfer that occurs at the grain boundaries, it facilitates intracrystalline plastic deformation in the form of dislocation creep, and this process is very rapid in halite (Schenk et al., 2006; Urai & Spiers, 2007; Urai et al., 1987). Diagenetic halite generally has a mosaic texture whereby crystal boundaries meet at triple junctions of nearly 120°, which might be facilitated by a pressure solution mechanism (Warren, 2006). Diffusive mass transfer can also occur at higher temperatures in the metamorphic domain, and then it occurs in the solid state. There are two main types of solid state diffusive mass transfer. Nabarro–Herring creep is caused by volume diffusion movement of lattice defects in the crystal lattice, and Cobble creep is the result of solid state diffusive mass transfer through grain boundaries triggered by chemical potential gradients linked to internal strain energy gradients.

Third, intracrystalline plastic behavior involves the movement of dislocations, and also solid state diffusion in some mechanisms. A dislocation is a defect or imperfection in the crystal lattice, and the movement of dislocations happens by breaking bonds in the original lattice ahead of the dislocation and reforming the bonds behind the dislocation. There are two main types of dislocation movement: dislocation glide by movement along slip planes, and dislocation climb by the change of slip plane of edge dislocations. Dislocation creep involves deformation by both dislocation movement types. Even at low confining pressures of 10 MPa, dislocation creep is important in rock salt in the diagenetic domain (Urai & Spiers, 2007). Dislocation creep causes several phenomena, namely (i) undulatory extinction due to the crystal lattice distortion, (ii) the formation of subgrains that have a slightly different crystallographic orientation than the rest of the grain, and (iii) the formation of new grains during dynamic recrystallization by subgrain rotation and grain boundary migration. Subgrain boundaries are oriented approximately perpendicular to the dislocation glide direction, and the subgrain diameter correlates with deviatoric stress (Carter et al., 1993). Intracrystalline plastic deformation causes flattened or elongated grains with a preferred orientation, and the movement of dislocation causes strain ellipsoids. Grain shape fabrics can also result from the coalescence of grains with similar orientation during dynamic recrystallization, with grain boundary migration driven by the internal strain energy gradient from a less-deformed grain with lower dislocation density to a more highly strained one (Ter Heege et al., 2005). High strain deformation experiments on relatively dry coarse-grained rock salt show that dislocation creep is the main microstructural deformation mechanism without extensive dynamic recrystallization (Linckens et al., 2016). Fluid-assisted grain boundary migration involves solution-precipitation transfer across grain boundary brine films, driven by differences in chemical potential between grain boundaries linked to dislocation density differences between old deformed grains and newly growing grains (Peach et al., 2001; Schenk et al., 2005, 2006). Fluid inclusions at the grain boundaries are transformed into a grain boundary fluid film during both static and dynamic recrystallization (Ghanbarzadeh et al., 2015). The relative importance of deformation mechanisms, such as dislocation creep and pressure solution, depends on temperature, confining pressure, grain size, solid solution impurities, second phase content, and water content.

Methods of microstructure identification

Microscopy is commonly used to document microstructures, and more advanced techniques are used for three-dimensional imaging of samples using X-ray computed tomography, microstructural damage evolution using acoustic emission method, and porosity distribution using nuclear magnetic resonance technique. Several types of sample preparation are applied to enhance microstructure visualization, and multiple types of microscopy are used such as polarization microscopy with transmitted light and incident light and scanning electron microscopy. Rock salt samples are particularly challenging in terms of preparation of thin sections for microscopic study, mainly because of the solubility of salt in water, the impact of humidity, and the effect of water content in the samples on microstructures and geomechanical properties. Therefore, rock slabs of salt samples need particular preparation by either dry sawing or preparation using oil.

Transmitted light microscopic analysis is conducted on thin sections and enables observation of fluid inclusions that are aligned at high angle grain boundaries. The inclusions can have various shapes, such as arrays of isolated bubbles, elongated bubbles or tubes, or even almost continuous films. Some sample treatment techniques have been documented that can further help microstructure identification. First, observation of grain and subgrain boundaries is facilitated by chemical polishing and etching of mechanically polished samples in 5.5 M NaCl solution with 0.8 wt% FeCl3 · 6H2O for 10 s, followed by rinsing the surface with n-hexane and drying with hot air (Urai et al., 1987). Incident light microscopy was used to study the etched salt sample surfaces. Such microscopic analysis enables distinction between high angle grain boundaries and subgrain boundaries by deeply etched and less deeply etched lines, respectively. Second, using the same etching solution but leaving it dry slowly on an etched sample surface to enable NaCl epitaxial overgrowth provides information on crystallographic orientation of salt grains (Friedman et al., 1984). Third, additional microstructural information can be obtained through irradiation of salt samples with gamma radiation with dose rates of 20 to 80 MRad/h for a total dose of around 5 × 109 Rad, and at a temperature of 120°C (Urai et al., 1985). Such irradiation causes a blue coloration of the samples due to induced lattice defects, which enables visualization of microstructural features (Levy et al., 1981).

A scanning electron microscope (SEM) is used to visualize microstructures and their morphologies (Martin-Clave et al., 2021). Cryogenic scanning electron microscopy (cryo-SEM) and environmental scanning electron microscopy (ESEM) allow the investigation of liquid or wet samples. This can be achieved by investigation of the sample in the frozen state in the case of cryo-SEM whereby small rock samples are subjected to shock freezing to approximately −190°C, or under reduced pressure in the case of ESEM. Specifically, ESEM is employed for the study of rock salt grain boundary healing. The use of ESEM enables the observation of water trapped at the surface of particles and meniscus formation between adjacent particles (Hwang et al., 1993). Cryo-SEM has also been used to study fluid-filled and healed grain boundaries by observation of the boundary fluids in the frozen state (Schenk et al., 2006). Similarly, the brine in grain boundaries is investigated using broad ion beam cryo-SEM to gain insights in the mechanism of static recrystallization of wet halite with the evolution of the grain boundary structure (Desbois et al., 2012). With this method, the broad ion beam is used to produce a cross section in the sample for further SEM investigation.

Electron backscatter diffraction (EBSD) analysis (Figure 5) is used to determine the crystallographic orientation of grain boundaries in rock salt samples (Schenk et al., 2006). For this type of analysis, the samples need to be very finely mechanically polished down to 4000-grade carborandum paper polishing, and 10 s chemical polishing with methanol, followed by diethyl ether rinsing (Schenk et al., 2006). Before EBSD analysis, the samples were coated with carbon to reduce charging. Analysis of EBSD patterns can be done using software through comparison with calculated halite patterns, and EBSD maps are further processed to remove erroneous data (Schenk et al., 2006).

X-ray computed tomography (CT) is a nondestructive test employed for three-dimensional visualization of microcracks in rocks by analysis of differences in density (Zhao, Ma, et al., 2022). Different minerals can be distinguished in the rock samples (especially for minerals with significant differences in density), and digital processing of the CT images enables quantification of pore and fracture space and different minerals. The CT images are segmented and brightness is adapted to enhance the contrast and improve the separation of different regions (rock matrix vs. pore space, and between different minerals). For example, anhydrite impurities, pore space, and fluid phases along grain boundaries, or intracrystalline fluid inclusions can be visualized and quantified using micro-CT (Thiemeyer et al., 2015). A CT scan of the rock salt sample before geomechanical testing serves as a control to ensure sample integrity (Zhao, Ma, et al., 2022). Micro-CT is also used to identify the changes in the internal microstructures upon mechanical loading the rock salt sample (Zhang, Agostini, et al., 2021). Moreover, the effect of the evolution of texture and internal stresses on the relationship between permeability and applied stress was studied by 3D synchrotron micro-CT and 3D X-ray diffraction on polycrystalline rock salt samples (Moslehy et al., 2021).

Acoustic emission (AE) is a common method applied to the studies on the evolution of damage and crack formation in rock samples. This method can be used for real-time monitoring, and can then be combined with post-test CT scanning of the rock sample. The AE technique is based on sound waves that are generated when microcracks are formed or are propagating (Zhao, Ma, et al., 2022). This is a result of the release of stored energy that builds up in the sample with stress concentrations at certain locations (especially near pores or already existing microcracks), due to external loads applied on the sample during mechanical testing. Analysis of the acoustic emission parameters including events, energy and frequency-amplitude distribution, allow for the reconstruction of the evolution of damage in time and space in the rock sample. Research on rock salt has demonstrated that the AE method can be used on plastic natural materials to investigate deformation under thermobaric conditions, and to study the effect of temperature and strain rate on fracture dynamics. Moreover, anomalous changes in AE activity allow the distinction between the different stages of rock salt deformation (Shkuratnik et al., 2020a). The AE method has also been used to investigate the stress memory effect in uniaxial cyclic loading of rock salt samples under variable temperatures (Shkuratnik et al., 2020b). Additional work explored experimentally the rock salt dilatancy boundary using AE and triaxial compression tests were carried out on rock salt samples from the Asse Salt Mine, Germany (Alkan et al., 2007). It is shown that the AE could be used to determine the dilatancy boundary during the triaxial tests. AE test systems can be further used to conduct experiments on bedded rock salt samples under uniaxial compression and indirect tension. The failure state of bedded rock salt can be predicted by inspecting the variation between the stress or the energy release, and the fractal dimension of the AE spatial distribution (Xie et al., 2011).

The nuclear magnetic resonance technique has been used on rock salt to investigate pore characteristics of healed rock salt samples after damage caused by uniaxial compression (Chen, Peng, et al., 2020), and to study changes in pore structure in rock salt subjected to cyclic loading (Wang, Zhang, et al., 2022). Nuclear magnetic resonance is the physical response of a nucleus that sits in a strong constant magnetic field and gets subjected to a weak oscillating magnetic field. Longitudinal and transverse relaxation times are defined as the mean time for the magnetic moment of the nucleus to get back to its original state upon disabling the oscillating magnetic field. Studies have derived a relationship between the transverse relaxation time T2 and the pore radius of a rock (Ge et al., 2014). NMR technology has also been employed to investigate water migration and gas desorption in pore-fracture systems using transverse relaxation time T2 (Zhang et al., 2019).

The pore structure of rock salt has also been studied using mercury injection porosimetry and gas adsorption testing (Chen et al., 2018; De Las Cuevas, 1997). The method of mercury injection is based on mercury, a nonwetting fluid, entering pores in samples when the pressure exceeds the capillary pressure. A pressure-volume curve is generated by gradually increasing the applied pressure. The pore size can be correlated to applied pressure using the Washburn equation. At very high pressure, the mercury injection method could lead to some damage in the sample. Therefore, another method, namely gas adsorption, is used in particular for determination of small pores of less than 100 nm in diameter (Zhang, Wei, Vandeginste, et al., 2020). This method is based on adsorption isotherms using an inert gas, generally nitrogen. The pore size is correlated to partial pressure. Several adsorption theories and models have been proposed for adsorption of differing gases in pores in various rock types, for example, CO2 in shale (Xie et al., 2022).

Rock salt physical and geomechanical properties

With low density that is, less than 2.3 g/cm3, low hardness that is, less than 3 on Mohs hardness scale, high solubility, low strength, rock salt typically deforms easily showing creep behavior (Giambastiani, 2020). It has a higher thermal conductivity (up to 6 W/m K) than other rocks such as sandstone, mudstone or coal (Mello et al., 1995; Zhuo et al., 2016). The viscosity (as measured by the ratio of shear stress to the shear strain rate) of rock salt is also lower than that of other rocks, and depends on the temperature and moisture content of the rock (Warren, 2017). Rock salt has low porosity and permeability, and good self-healing properties, making it an desirable medium for underground gas storage (Liu & Xiao, 2014). Halite has a density of 2.17 g/cm3 at room temperature, which decreases with greater depth due to greater thermal expansion and insufficient porosity to accommodate the expansion. For example, at a temperature of 150°C and a depth of 5 km, halite shows an expansion of 2% caused by heat and contraction of 0.5% due to pressure (Talbot & Jackson, 1987). Moreover, in comparison to other subsurface sedimentary rocks that lose porosity and increase in density with deeper burial, halite becomes less dense than surrounding rocks from about 1 km depth, which drives positive buoyancy of rock salt. An overview of the density and Mohs hardness of some of the most common minerals in rock salt (Giambastiani, 2020) is presented in Table 1. All minerals in rock salt are quite light, with the lightest being carnallite and sylvite, whereas anhydrite and polyhalite are the heaviest. Gypsum and sylvite are the softest from the series, and polyhalite and anhydrite the hardest.

The geomechanical properties based on short-term mechanical tests are presented in Table 2. This overview is based on the studies and references compiled by Giambastiani (2020). Anhydrite is not only the hardest in the series of common rock salt minerals, anhydrite rocks are also significantly stronger than gypsum rocks and halite rocks (Bell, 1981). The average E-modulus of anhydrite rocks is also twice as high as that of halite rocks and gypsum rocks (Table 2). Moreover, the common rock salt types all show plastic deformation before failure, and anhydrite rocks show the least and halite rocks the most amongst a comparison between halite, anhydrite, gypsum, and sylvite rocks (Bell, 1981).

The rheological behavior of rock salt depends on external factors, such as applied loading, confining pressure, and temperature, as well as intrinsic properties, including mineralogical composition, water content, texture, grain size, and impurities. Two deformation domains, namely a dilatancy domain and a compaction domain, have been established for rock salt (Cristescu & Hunsche, 1998; Van Sambeek et al., 1993); these are separated by the dilatancy boundary (Figure 6). The latter depends on pore pressure and stress loading rates (Alkan et al., 2007). Dilation is faster with higher pore pressure, and the dilatancy boundary values decrease slightly with faster stress loading (Alkan et al., 2007). Dilation, which is caused by microfracturing, results in an increase in rock salt permeability (Van Sambeek et al., 1993). This brittle deformation occurs at very low effective confining pressure and high deviatoric stress (Cristescu & Hunsche, 1998; Urai et al., 2008), and leads to damage of the rock structure. Dilation is higher with a higher axial pressure, and faster dilation occurs for slower unloading and at higher temperature (Zhao et al., 2019). Dilatation or volume growth is of particular concern for the application of underground storage of gas or hazardous waste. In the compaction domain, deformation occurs in a ductile manner without the formation of cracks or failures, and with salt deformation by creep due to movement of dislocations and elastic deformation (Hunsche & Hampel, 1999). Such creep deformation is time-dependent. It is a very important phenomenon in rock salt, where low stress results in large strain without brittle failure. Macroscopic geomechanical behavior can be linked to microstructural mechanisms. During dislocation creep, subgrains are formed, and fluid assisted grain boundary migration may occur if water is present in the form of brine inclusions or grain boundary films (Schenk et al., 2005). Solution-precipitation creep may take place alongside intergranular sliding and rotation.

The creep behavior of rock salt can be split into three domains based on the change in strain with time, namely, primary or transient creep, secondary or steady state creep and tertiary or accelerated creep (Figure 7). During primary creep, the strain increases very rapidly at the beginning, and then stabilizes gradually, leading to secondary creep where the strain rate is constant. Tertiary creep is the final stage of deformation before failure, when the strain rate increases exponentially until failure. Also, under combined creep and fatigue, the strain-time curves show a similar pattern with the three domains of transient, steady state and tertiary creep (Ma, Wang, Wang, et al., 2021). The steady state strain rate (considering nondilatant deformation) can be expressed as the sum of the strain rate caused by dislocation creep, and that caused by pressure solution creep, which are functions of the respective activation energy, temperature, and differential stress:

(2)

where A and B denote constants related to deformation rate (s −1); Q d is the activation energy for dislocation creep (J/mol); Q p is the activation energy for pressure solution creep (J/mol); R is the universal gas constant (J/mol K); T is the temperature (K); σ is the differential stress (MPa); D is the grain size; n is the exponential factor for stress and m is the exponential factor for grain size (Urai & Spiers, 2007). The correlations reflect that dislocation creep is strongly dependent on differential stress and pressure solution creep is heavily dependent on grain size. Moreover, the creep behavior of interbedded argillaceous rock salt is different from that of high purity rock salt (Gao et al., 2022).

Creep has not only been studied under single-stage loading, but also with multistage loading. Results have indicated that the crack development is more extensive in samples that have experienced multistage loading in comparison with single-stage loading, and this damage accumulation leads to deterioration of the rock salt geomechanical properties (Dong et al., 2022). The effect of this damage accumulation is mainly reflected in the transient strain and the strain in the transient creep domain, whereas the steady-state creep is related to stress and temperature but independent of loading history (Dong et al., 2022). The long-term strength of rock salt was determined and a nonlinear creep damage constitutive model was established by Wu et al. (2020) based on multistage creep tests.

Geomechanical testing methods

Rock salt geomechanical testing can be conducted using short term or long term experiments, uniaxial or triaxial compression testing, and with static loading or cyclic loading. The uniaxial compression test uses compression along the length axis of the sample whereas the sample is unconfined in the transversal direction. Under triaxial compression testing, compression loads are applied both in the longitudinal direction and in the transversal direction. Generally, cylindrical rock salt samples with a diameter of at least 10 times the crystal size, and a length-to-diameter ratio of two (Hawkins, 1998), are placed between two steel platens and a layer of heat resistant silicon tape. In particular for the triaxial experiments, the rock salt samples are also covered by heat shrink PTFE membrane to separate the sample from the confining fluid (Martin-Clave et al., 2021). Axial extensometers and circumferential chain extensometers are attached to the sample to measure dimension changes, and thus to derive strain and Poisson ratio. The force transducer (load cell) enables measurement of the load applied to the sample.

Geomechanical tests can apply static loads, meaning a constant load in the longitudinal direction, and in the case of triaxial loading also a load in the transversal direction, for the entire duration of the experiment. By calculating the stresses and strains based on the measurements during the experiment, the elasticity modulus and Poisson ratio can be derived. Triaxial compression tests have also been combined with acoustic emission measurements (Figure 8) to determine the dilatancy boundary under a specific stress, stress loading rate and pore pressure (Alkan et al., 2007). Cyclic loading is applied to test the fatigue strength and behavior of the material (Wang, Zhang, et al., 2022). In cyclic loading geomechanical tests, the cycling rate and the number of cycles can be set. In general, cyclic loading of a material would be performed until failure to determine the fatigue strength at a certain maximum stress and amplitude.

Besides short-term geomechanical tests that are used to determine rock salt properties, creep tests are conducted, as time-dependent or long-term geomechanical behavior is very important for rock salt (Eslami Andargoli et al., 2019). The strength of rock salt is dependent on the strain rate. Rock salt can fail in a brittle manner through the formation of fractures under high strain rates, but subsurface rock salt tends to flow (Talbot & Jackson, 1987). Creep tests apply static loading and are conducted over longer periods of time, varying from at least several months to years (Lyu et al., 2021). Very slow creep tests (with stages of 8 months) have been conducted on natural salt samples under a small deviatoric stress range to assess the applicability of constitutive laws that are extrapolated from relatively high stress tests (Bérest et al., 2019). The results showed that transient creep takes long (6 to 10 months) and steady-state strain rates are much faster (7 to 8 orders of magnitude faster) for these tests with small deviatoric stress in comparison to that predicted by most constitutive laws (Bérest et al., 2019). Furthermore, time-dependent behavior has also been assessed using cyclic loading tests, whereby the effect of loading frequency was evaluated and led to establishing a variable-parameter creep damage model (Ma, Wang, Wang, et al., 2021). Another study using a similar method indicated a positive interaction between the time-dependent damage and the cumulative cyclic damage in rock salt (Ma, Wang, Wang, et al., 2021). The study suggested that the life of the rock salt can be predicted through the linear correlation between the logarithm of the deformation rate in the steady-state phase and the logarithm of the creep or fatigue life, that is independent of the loading history (Ma, Wang, Wang, et al., 2021).

Rock salt cavern construction methods

Caverns can be made artificially in rock salt by salt solution mining, which involves dissolving salt with water and pumping the brine back up to the surface (Figure 11). For caverns in halite, one cubic meter of rock can be dissolved with seven to eight cubic meters of fresh water pumped into the cavern. Accelerated rock salt leaching is achieved by the use of high-pressure water jets with water stream pressure of 500 bar (Korzeniowski et al., 2020). The rock salt dissolution in dynamic water and the fluid transport mechanism during solution mining have been modeled based on kinetics of the convection-diffusion process and the theory of fluid dynamics and chemical kinetics (Yang & Liu, 2017). In the solution mining under gas method for the construction of salt caverns, a protective fluid (blanket), that is insoluble in water, lighter than water and unreactive with water or salt) is used to protect the casing shoe and to control the cavern shape (Sedaee et al., 2019). Commonly, liquid hydrocarbons are used as blankets (Favret, 2004), but due to their cost and environmental concerns, nitrogen gas has been studied for its use as a blanket for the construction of a rock salt cavern for underground gas storage, and optimized leaching parameters can be predicted with simulation of water injection and nitrogen gas injection pressure (Wang et al., 2021). In bedded salt, where the caverns have greater horizontal dimensions, solution mining with double-well convection is employed (Zhang, Wang, et al., 2020).

Rock salt mineralogy and cavern construction

Salt solution mining generally targets thick rather homogeneous halite formations. Rock salt existing in diapirs is therefore preferred over bedded salt formations. The composition and heterogeneity of the rock salt is important, since differences in dissolution behavior will influence the shape of the cavern, and cause potential collapse of less soluble components, such as anhydrite or dolomite. Moreover, as is shown by the studies on underground gas storage in rock salt caverns in China, the bedded rock salt nature with interlayers of anhydrite, glauberite and mudstone leads to collapse and piling up of insoluble mudstone at the bottom of the cavern during salt solution mining, thus causing a one-third or even two thirds reduction of the gas storage capacity of the cavern (Li et al., 2016). Moreover, the mudstones are volumetrically expanded after collapse in the cavern due to clay swelling, an increase in the volume of the pore space free water, and water bound to the surface of the particles (Yao et al., 2022). Apart from the geological characteristics, the construction technology and tubing failure are also considered as factors that can lead to the formation of irregular caverns, and their operation comes with risks and safety hazards (Xue et al., 2020). Furthermore, solution mining simulation models have shown that creep deformation induced volumetric shrinkage and displacement of the cavern roof stabilizes at a length of 600 m (or at a length-to-diameter ratio of around 5 to 6) for horizontal cavern structures in bedded salt formations (Li, Zhang, et al., 2022).

Rock salt cavern depth, shape, and microstructures

Generally, rock salt caverns are made through solution mining in formations at depths between around 200 and 2000 m. Greater depths would lead to significant creep and cavern closure, as a larger pressure difference causes a greater cavern shrinkage rate (Zhang, Wang, et al., 2018). The rate of cavern closure is strongly dependent on grain size, with much higher rates for fine grained rock salt caused predominantly by pressure solution creep (Cornet et al., 2018). Caverns at shallower depths may be prone to catastrophic failure. Models have shown that collapse depth does not increase linearly with the burial depth. Therefore, collapses of rock salt caverns generally happen at depths shallower than 600 m (Zhang, Wang, et al., 2018). In rock salt layers at depths of more than 900 m, there are common issues with shaft lining deformation, and recently, a new methodology involving local and controlled leaching to remove the excessive creep into the outbreak side of the shaft diameter has been proposed (Kaminski, 2021). Besides the cavern depth, the pressure and temperature gradients of the overburden, the pressure inside the cavern, the shape of the cavern and the rock salt properties such as grain size and moisture content will also influence the rate of creep deformation. Caverns with irregular shaped walls show a lower shrinkage in volume and less displacement of the cavern walls (in comparison to regular shaped salt caverns). Larger plastic deformation in these caverns is located to their overhanging and concave parts (Liu, Zhang, Fan, Jiang, et al., 2020). Numerical simulations have indicated that ellipsoid shaped caverns are more stable than cylindrical or cuboid shaped caverns, and that the roof shape (which should be designed as an arch) is more important than the shape of the side walls with regard to the stability of the caverns (Liu, Zhang, Fan, Jiang, et al., 2020). Physical simulation technologies and testing systems of single-well solution mining in rock salt have been developed for cavern shape control (Wang, Wang, et al., 2022).

Geomechanical properties of rock salt caverns for gas storage

Beyond the purpose of the production of salt from salt solution mining, the rock salt caverns can serve as storage containers for hydrocarbons and waste. Moreover, rock salt caverns have gained great attention recently in the renewable energy field as potential storage sites for hydrogen (AbuAisha & Billiotte, 2021) or compressed air (Khaledi et al., 2016). A brief overview of the history of storage of fluids and gases in underground salt caverns is presented in Firme et al. (2019). The development of an underground cavity results in a response of the surrounding rock, which expands into the newly created void volume, and thus inelastic dilatation (volume expansion) occurs. The latter triggers a decrease in fluid pressure, which may result in the generation of a gas-fluid (two-phase) system (Beauheim & Roberts, 2002). With small amounts of dilation (≥0.05 vol%) sufficient microfractures form to create an interconnected pore network beyond the percolation threshold (Peach, 1991). Hence, dilation around an underground void volume can cause halite to shift from practically nonexistent permeability to measurable permeability. Moreover, dilatancy, and thus permeability, is enhanced by second phase impurities in the halite such as anhydrite and clay (Beauheim & Roberts, 2002). Permeability, which is a key factor in predicting the integrity of the storage site, is not simply a function of stresses or strains. It is a very nonlinear function of dilatancy, impacted by microcrack linkage (Popp et al., 2001), and it is strongly dependent on the type of salt. Nevertheless, the dilatancy boundary is considered a safety boundary for permeability and integrity (Hunsche & Hampel, 1999).

In the context of renewable energy storage, rock salt caverns provide a solution to deal swiftly with peak demand and demand fluctuations. The caverns enable rapid cycling with a 15 min's withdrawal to injection changeover time. Such caverns for storage are generally at a depth of 1 to 2 km below the surface, and about 110 m below the top of the salt formation. It is best to use rock salt caverns in diapirs for gas storage rather than bedded rock salt, since the interface between different lithologies may be a potential leakage site, and a focus location for strain weakening, for example, weak interfaces between rock salt and mudstone beds (Zhang et al., 2014). Injection and withdrawal of gas results in pressure cycling. Large stress cycles can lead to inelastic deformation around salt cavern walls and trigger the development of grain boundary cracking and frictional sliding enabling the redistribution of water and potentially initiating viscous processes (Ding et al., 2021). Instability of cavern walls could be caused by the production of linked arrays through the linkage of preferentially opened grain boundaries with shear slipping boundaries (Ding et al., 2017). Large stress cycling should thus be avoided, and holds could be employed to allow stress relaxation and healing of cracked grain boundaries (Ding et al., 2021). The temperature will also be affected by repeated injections and withdrawal of gas in rock salt caverns. The injected compressed gas is generally cooler than the temperature of the rock salt enclosing the cavern. Moreover, a rapid gas pressure drop in the cavern results in cooling, and thus thermal contraction of the rock salt, causing tensile stress (Bérest et al., 2007). Studies on high frequency cycling of salt caverns have indicated that thermally induced tensile fracturing is possible, and thermo-mechanical modeling can help predict the location, orientation, and timing of the first fractures (Blanco-Martín et al., 2018). Studies on permeability evolution in rock salt have shown that dynamic mechanical and thermal fatigue may lead to a slight increase in permeability due to microcracking and a reduction in permeability during static fatigue (creep) due to self-healing (Grgic et al., 2022). The mechanisms behind self-healing of damage in rock salt involve mechanical closure, diffusive healing driven by a reduction in surface energy, and healing by recrystallization; and the influencing factors include stress, temperature, initial damage, humidity, and chemical environment (Kang et al., 2019). The effects of the tempo-spatial variation of temperature and pressure (Figure 12) on the mechanical behavior of the cavern have also been simulated using a coupled aero-thermo-mechanical model (Li et al., 2023). Monitoring of acoustic emissions was also employed to investigate the impact of salt cooling related to fast injection and withdrawal cycles. Studies have shown that the first cooling period displays the strongest and deepest acoustic emissions (Balland et al., 2018). If the cavern is not used for a long time, the temperature inside the cavern will increase, resulting in higher pressure.