The E ﬀ ect of Weathering on Salt Release from Coal Mine Spoils

: Coal mine spoils have the potential to create environmental impacts, such as salt load to surrounding environments, particularly when exposed to weathering processes. This study was conducted to understand the e ﬀ ect of physical and chemical weathering on the magnitude, rate, and dynamics of salt release from di ﬀ erent coal mine spoils. Five spoil samples from three mines in Queensland were sieved to three di ﬀ erent particle size fractions ( < 2 mm, 2–6 mm, and > 6 mm). Two samples were dispersive spoils, and three samples were nondispersive spoils. The spoils were subjected to seven wet–dry cycles, where the samples were periodically leached with deionised water. The rate, magnitude, and dynamics of solutes released from spoils were spoil speciﬁc. One set of spoils did not show any evidence of weathering, but initially had higher accumulation of salts. In contrast, broad oxidative weathering occurred in another set of spoils; this led to acid generation and resulted in physical weathering, promoting adsorption–desorption and dissolution and, thus, a greater release of salts. This study indicated that the rate and magnitude of salt release decreased with increasing particle size. Nevertheless, when the spoil is dispersive, the degree of weathering manages salt release irrespective of initial particle size. This study revealed that the long-term salt release from spoils is not only governed by geochemistry, weathering degree, and particle size but also controlled by the water / rock ratio and hydrological conditions of spoils.


Introduction
During coal mining activities, overburden is removed and deposited on land in the form of waste rock dumps commonly known as spoil piles [1][2][3]. When these coal mine spoils are exposed to atmospheric conditions, weathering processes occur that can potentially generate adverse environmental impacts through salt release into final voids and water ways, and/or salt crust formation on the land surface restricting plant growth [4,5]. The potential for impacts can be managed by rehabilitation, for example, by using soil and vegetation covers that exclude most of the rainfall from the underlying rock [2]; however, prior to, during, and following rehabilitation, it is essential to understand the trajectory and magnitude of salt release and transport from spoils.
Salt generation from a spoil heap typically depends on the geochemical properties of the rock [1,6]. As an example, in Australia, water seepage from coal mine spoils tends to be saline due to the presence conditions for 28 days. The ambient potential evaporation ranged between 0.22 and 9.36 mm day −1 , and the temperature of the environment fluctuated between 22.8 and 31.4 • C.
The degradation test indicated that some spoils can physically break down to smaller particles (<2 mm), while no physical changes were observed in other spoils over the period of the experiment (12 months). Based on the test, we classified the studied coal mine spoils into two groups-soil-like spoils (dispersive spoils: spoils with high degradation degree) and rock-like spoils (nondispersive spoils: spoils with low degradation degree)-to assess the effect of weathering on their salt release potential. This terminology (soil-like and rock-like) was also consistent with the geotechnical classification of spoils [39]. A comprehensive description of the degradation study was reported in Edraki et al. [37].

Funnel Leaching Experiment
A series of leaching experiments were conducted using free-draining funnels to simulate the weathering conditions on the studied spoils using wet-dry cycles. The spoil samples were firstly sieved to three different fractions: <2 mm, 2-6 mm, and >6 mm. Therefore, based on the spoils' type, classification (i.e., rock-like or soil-like) and size, the studied spoils were named as follows: A10RL2 (A10, rock-like spoil with <2 mm particles); A10RL26 (A10, rock-like spoil with 2-6 mm particles); A10RL6 (A10, rock-like spoil with >6 mm particles); B8RL2 (B8, rock-like spoil with <2 mm particles); B8RL26 (B8, rock-like spoil with 2-6 mm particles); B8RL6 (B8, rock-like spoil with >6 mm particles); B9SL2 (B9, soil-like spoil with <2 mm particles); B9SL26 (B9, soil-like spoil with 2-6 mm particles); B9SL6 (B9, soil-like spoil with >6 mm particles); C4RL2 (C4, rock-like spoil with <2 mm particles); C4RL26 (C4, rock-like spoil with 2-6 mm particles); C4RL6 (C4, rock-like spoil with >6 mm particles); C6SL2 (C6, soil-like spoil with <2 mm particles); C6SL26 (C6, soil-like spoil with 2-6 mm particles); C6SL6 (C6, soil-like spoil with >6 mm particles). Buchner funnels (diameter: 120 mm; height: 60 mm) were filled with the spoils of different sizes with bulk density of between 0.86 and 1.47 g cm −3 depending on the spoil type and particle size class ( Table 2). The selected particle size classes were the particle sizes that could be adapted to the small size of the Buchner funnels used for the experiment. No replicate was produced for each funnel as funnels were representative of the spoil piles and producing an equal bulk density for the funnels filled with the same spoils was not achievable due to the particle size heterogeneity of the spoils.
The funnels were subjected to seven wet-dry cycles by exposure to atmospheric conditions (ambient oxygen, temperature, humidity, and evaporation) following leaching by DI water. During wetting cycles, the funnels were leached with 550 mL of DI water. This is equivalent to a 49 mm depth of water, representing a typical intensive rainfall event in the Bowen Basin. The DI water was slowly applied to the funnels to prevent surface disturbance. After each wet cycle, the funnels were exposed to atmospheric laboratory conditions for approximately 30 days. The potential evaporation in the laboratory ranged between 0.57 and 1.20 mm day −1 , and the temperature of the environment fluctuated between 22 and 26 • C.
During each wet cycle, ten successive leachate solutions were collected, each containing one-tenth of the total leachate volume collected. Each leachate solution was then analysed for EC (electrical conductivity) and pH at 20 • C using a TPS 901-CP meter (TPS Pty Ltd, Brisbane, Australia). To analyse the leachate solutions for the major cations, chloride, and total S, we followed the Park et al. [6] method. From Cycles 1 to 3, we combined each three leachate solutions (150 mL) of each leaching cycle and discarded the tenth leachate solution [6]. From Cycles 4 to 7, we combined the first three leachate solutions (150 mL) and discharged the remaining leachate solutions [6]. Subsequently, the major cations and total S were analysed using ICP-OES (inductively coupled plasma optical emission spectrometry) (ThermoFisher Scientific, Waltham, MA, USA). Chloride concentrations were also measured using IC (ion chromatography) (ThermoFisher Scientific, Waltham, MA, USA).

Chemical Analyses
To measure the concentration of soluble salts prior to any leaching and after the final wet-dry cycle (at the end of the experiment), samples were pulverised to <2 mm. The soluble salts were then analysed using a 1:5 (w (weight)/v (volume)) extraction (pulverised spoils/DI water) by mixing for 2 h, resting for 2 h, and then centrifuging (5 min at 3000 rpm) before collecting and filtering the supernatants (using 0.045 µm filter) to analyse the major soluble cations (Na + , K + , Mg 2+ , Ca 2+ ) by ICP-OES [40]. The exchangeable cations of the spoils were measured using a 1:20 w/v (pulverised spoils/1 M NH 4 Cl) extraction by mixing for 2 h and resting for 2 h, followed by centrifugation (5 min at 3000 rpm) and filtering (using 0.045 µm filter) [40]. Before analysing the samples for exchangeable cations, the spoils were treated with 60% ethanol and 20% glycerol to eliminate soluble salts [40].
A 1:5 (w/v) suspension of pulverised spoils with DI water was mixed for 1 h and left to stand for 1 h before measuring the supernatant for EC and pH at 20 • C using a TPS 901-CP meter [40]. To measure the chemical composition of spoils, a small sample (250 mg) of each pulverised spoil (<2 mm) was collected and digested using 5 mL nitric acid (HNO 3 ; 70%), 3 mL hydrochloric acid (HCl; 32%), and 2 mL hydrofluoric acid (HF; 50%) for 1 h at room temperature (22 • C). For further digestion, the samples were placed in the carousel in an Ethos-1 digester and heated at 100% power for 40 min at 200 • C. After digestion, the samples were transferred to a 25 mL volumetric flask and the sample was topped up with DI water. Solution samples were then collected to determine the chemical components using ICP-OES. The mineralogy of the spoils was also determined using the X-ray diffraction method. The X-ray diffraction pattern was produced using a Bruker-AXS D4 XRD (Bruker, Karlsruhe, Germany) with copper radiation at 30 kV and 10 mA. The powder mounts were run over a 2θ beam range of 4 to 80 • , with a 0.02 degree step and a 2 s per step count time.

Physical Analyses
The flow rate of the first leachate collected at each cycle was calculated based on the volume of leachate solution, the time taken to collect it, and the cross-sectional area [41]. The bulk density of the samples was calculated based on the following equation [42]. The particle density values of the spoils were measured using a gas pycnometer [41]. Consequently, the porosity of each spoil was calculated as follows [41]: The particle size distributions of three samples (C6SL spoils) were analysed using the hydrometer method as described in Sheldrick and Wang [43]. A 10 mL volume of sodium hexametaphosphate solution (10%) was mixed with each spoil sample (25 g), and a sonifier (Branson model 450) was used for 10 min to break the aggregates [43]. The water retention characteristics of these samples (C6SL spoils) were also determined by the Rosetta model using HYDRUS [44,45]. To calculate the macropore volume of each sample, the water potentials at field capacity were firstly determined based on the following equation as described in Assouline and Or [46].
Subsequently, the volumetric water contents at field capacity were calculated based on the van Genuchten model [47]: where θ (Ψ) is the water content (L 3 L −3 ); |Ψ| is the suction pressure (L); θ s is the saturated water content (L 3 L −3 ); θ r is the residual water content (L 3 L −3 ); α is the inverse of the air entry suction (1/L); and n is the measure of the pore size distribution. The macropore volume was then calculated based on the volumetric water content at saturation and field capacity. We analysed the particle size distribution and water retention characteristics for C6SL spoils only, because these spoils were the only ones that showed physical changes (breaking down of the particles) during the laboratory funnel leaching experiment ( Figure A1).

Dynamics of Salt Release and Salt Production Rate
To evaluate the effect of weathering and particle size on the dynamics of salt release from the studied spoils, breakthrough curves were presented with parameters (chemistry of leachate solutions) plotted against the leached pore volume. The studied spoils had different porosity values that resulted in a different number of leached pore volumes for each sample. The same approach was used by other studies [8,9,48] which evaluated the kinetics and the processes involved in salt release from porous media. As the composition of applied water remained constant during the experiment, the breakthrough curves can show detailed variations in the cation composition of the leachate, reflecting processes that occur in the materials over the experiment [48]. In order to be able to use the laboratory data for the assessment of salt release rates from spoils, the mass of salt generated per unit of time was calculated using the following equation [49].
Salt production rate = Solute concentration × Volume o f leachate collected Mass o f dry material × leach time Spearman's correlation coefficients (p < 0.05) were calculated using R software (Version 3.3.3).

Changes in Physical Properties of the Coal Mine Spoils during Weathering
The initial porosities of samples, which depended on the spoil type and particle size, were found to be between 0.43 and 0.67 cm 3 cm −3 ( Table 2). The average flow rate increased with increasing particle size for all spoils, excluding the B9SL26 sample, which had a lower flow rate (0.64 mm s −1 ) compared with the other B9SL spoils ( Table 2). The flow rates for C6SL spoils with >2 mm particles decreased over the first three cycles (20 pore volumes of leaching) (Figure 1e), while no such changes in flow rates were observed for the other spoils of that particle size class ( Figure 1). A broad physical breakdown occurred throughout the column after the third and second wet cycle for the C6SL26 and C6SL6 spoils, respectively ( Figure A1). These spoils broke down to smaller particles with approximately 23.7% clay content (Table 3). A reduction (up to 13%) was also observed in the porosity values of the C6SL samples (Table 3). A large permanent crack was observed in the C6SL2 sample at the end of the first drying period before the second wet cycle ( Figure A1). In contrast, other samples did not show any apparent evidence of physical changes (i.e., permanent vertical cracking) during the experiment ( Figure A1).  : inverse of the air entry suction; n: measure of the pore size distribution; Θs: saturated water content (porosity); Θr: residual water content.

Changes in the Chemistry of Leachate Solutions during Weathering
The EC of early leachate solutions was greater (22.1 dS m −1 ) for the A10RL2 sample than for other spoils ( Figure 2). The changes in the EC of the leachate solutions for all A10RL samples followed similar patterns, and decreased gradually to less than 0.58 dS m −1 during leaching of 80 pore volumes (over seven cycles) (Figure 2a). There were no major differences between the ECs of early leachate solutions for B8RL and B9SL spoils with the same particle size class (Figure 2b,c). Over the course of the experiment, the EC of leachate solutions was reduced slightly for B8RL and B9SL spoils with <6 mm particles. However, within each wet cycle, an increase in salinity was observed for the B8RL26 sample (Figure 2b,c). The leachate salinity for the B8RL2 sample also intensified during the sixth wet cycle before reducing to its initial extent ( Figure 2b). The EC of the leachate was relatively stable for the B8RL6 and B9SL6 samples (Figure 2b,c). A broad physical breakdown occurred throughout the column after the third and second wet cycle for the C6SL26 and C6SL6 spoils, respectively ( Figure A1). These spoils broke down to smaller particles with approximately 23.7% clay content (Table 3). A reduction (up to 13%) was also observed in the porosity values of the C6SL samples (Table 3). A large permanent crack was observed in the C6SL2 sample at the end of the first drying period before the second wet cycle ( Figure A1). In contrast, other samples did not show any apparent evidence of physical changes (i.e., permanent vertical cracking) during the experiment ( Figure A1). Table 3. Particle size distribution and water retention characteristics of the C6SL spoils at the end of the experiment. α: inverse of the air entry suction; n: measure of the pore size distribution; Θ s : saturated water content (porosity); Θ r : residual water content.

Changes in the Chemistry of Leachate Solutions during Weathering
The EC of early leachate solutions was greater (22.1 dS m −1 ) for the A10RL2 sample than for other spoils ( Figure 2). The changes in the EC of the leachate solutions for all A10RL samples followed similar patterns, and decreased gradually to less than 0.58 dS m −1 during leaching of 80 pore volumes (over seven cycles) (Figure 2a). There were no major differences between the ECs of early leachate solutions for B8RL and B9SL spoils with the same particle size class (Figure 2b,c). Over the course of the experiment, the EC of leachate solutions was reduced slightly for B8RL and B9SL spoils with <6 mm particles. However, within each wet cycle, an increase in salinity was observed for the B8RL26 sample (Figure 2b,c). The leachate salinity for the B8RL2 sample also intensified during the sixth wet cycle before reducing to its initial extent (Figure 2b). The EC of the leachate was relatively stable for the B8RL6 and B9SL6 samples (Figure 2b,c).
approaching steady state conditions (Figure 2d). The leachate EC from C4RL samples with >2 mm particles declined slightly with some cyclic increases in salinity at the beginning of each cycle ( Figure  2d). The EC for the C6SL2 sample reduced from 22 dS m −1 to less than 1.7 dS m −1 during the first cycle (nine pore volumes of leaching) before a substantial decline to 0.2 dS m −1 at the beginning of the second cycle (Figure 2e). The EC then increased instantly to 1.7 dS m −1 before decreasing and remaining relatively constant (Figure 2e). The leachate EC from the C6SL26 sample fluctuated during the first three cycles before increasing to 5.8 dS m −1 and then decreasing to less than 0.2 dS m −1 ( Figure  2e). A similar pattern was also observed for the C6SL6 sample with variations in the first two cycles before reducing constantly to 0.3 dS m −1 (Figure 2e).
The pH values for leachate solutions from the C6SL spoils ranged between 2 and 4 ( Figure A2). Conversely, other studied spoils (A10RL, B8RL, B9SL, and C4RL) released alkaline leachate solutions, with their pH values ranging between 6 and 10 depending on the spoil type and size ( Figure A2). The initial Na + concentration of the leachate was greater for the A10RL2 sample (165.8 mmol L −1 ) than for the other studied spoils ( Figure 3). Subsequently, the greatest reduction (164.2 mmol L −1 ) in Na + discharge was observed from this sample ( Figure 3). The release of Na + from the A10RL, B8RL, and B9SL spoils followed similar patterns, and Na + release decreased over the experiment (Figure 3). The sodium ion concentrations of leachate solutions for the C4RL2 sample initially increased from 45 mmol L −1 to 56 mmol L −1 before decreasing gradually to less than 0.6 mmol L −1 over 50 pore volumes (Figure 3d). A slight reduction was also observed in the concentrations of Na + from C4RL spoils with larger particle sizes (>2 mm), with Na + falling to less than 1.64 and 1.83 mmol L −1 for C4RL26 and C4RL6, respectively, at the end of the experiment (Figure 3d). For the C6SL2 sample, Na + concentrations sharply reduced from 0.47 to 0.12 mmol L −1 at the second cycle (14 pore volumes) before increasing to 0.28 mmol L −1 at the end of the experiment (Figure 3e). Likewise, the Na + concentrations decreased until the third cycle (22 pore volumes) for C6SL samples with >2 mm particles before increasing to approximately their initial extents at the end of the experiment ( Figure  3e). The continuous application of DI water to C4RL2 resulted in an initial rapid increase in EC, up to 8.3 dS m −1 during the first cycle (up to three pore volumes) (Figure 2d). The EC then gradually reduced to less than 0.3 dS m −1 after 25 pore volumes of leaching (the third wet cycle) before approaching steady state conditions (Figure 2d). The leachate EC from C4RL samples with >2 mm particles declined slightly with some cyclic increases in salinity at the beginning of each cycle (Figure 2d). The EC for the C6SL2 sample reduced from 22 dS m −1 to less than 1.7 dS m −1 during the first cycle (nine pore volumes of leaching) before a substantial decline to 0.2 dS m −1 at the beginning of the second cycle ( Figure 2e). The EC then increased instantly to 1.7 dS m −1 before decreasing and remaining relatively constant (Figure 2e). The leachate EC from the C6SL26 sample fluctuated during the first three cycles before increasing to 5.8 dS m −1 and then decreasing to less than 0.2 dS m −1 (Figure 2e). A similar pattern was also observed for the C6SL6 sample with variations in the first two cycles before reducing constantly to 0.3 dS m −1 (Figure 2e).
The pH values for leachate solutions from the C6SL spoils ranged between 2 and 4 ( Figure A2). Conversely, other studied spoils (A10RL, B8RL, B9SL, and C4RL) released alkaline leachate solutions, with their pH values ranging between 6 and 10 depending on the spoil type and size ( Figure A2).
The initial Na + concentration of the leachate was greater for the A10RL2 sample (165.8 mmol L −1 ) than for the other studied spoils (Figure 3). Subsequently, the greatest reduction (164.2 mmol L −1 ) in Na + discharge was observed from this sample (Figure 3). The release of Na + from the A10RL, B8RL, and B9SL spoils followed similar patterns, and Na + release decreased over the experiment (Figure 3). The sodium ion concentrations of leachate solutions for the C4RL2 sample initially increased from 45 mmol L −1 to 56 mmol L −1 before decreasing gradually to less than 0.6 mmol L −1 over 50 pore volumes (Figure 3d). A slight reduction was also observed in the concentrations of Na + from C4RL spoils with larger particle sizes (>2 mm), with Na + falling to less than 1.64 and 1.83 mmol L −1 for Minerals 2019, 9, 760 9 of 23 C4RL26 and C4RL6, respectively, at the end of the experiment (Figure 3d). For the C6SL2 sample, Na + concentrations sharply reduced from 0.47 to 0.12 mmol L −1 at the second cycle (14 pore volumes) before increasing to 0.28 mmol L −1 at the end of the experiment (Figure 3e). Likewise, the Na + concentrations decreased until the third cycle (22 pore volumes) for C6SL samples with >2 mm particles before increasing to approximately their initial extents at the end of the experiment (Figure 3e).
The dynamics of Ca 2+ release from the A10RL, B8RL, B9SL, C4RL, and C6SL2 spoils followed a similar pattern to the release of Na + from these spoils (Figures 3 and 4). In contrast, Ca 2+ concentrations in the leachate solutions from the C6SL26 sample increased to 1.4 mmol L −1 at the fourth cycle, before decreasing to 0.08 mmol L −1 at the end of the experiment (Figure 4e). Calcium ion release from the C6SL6 sample followed a similar trend, but the maximum generation of Ca 2+ occurred at the third cycle (Figure 4e).
The release of Mg 2+ correlated (ρ > 0.70, p < 0.05) with the release of Ca 2+ for all spoils, excluding the B8RL26 and C4RL6 spoils ( Figure A3). The trajectory of K + release was spoil specific ( Figure 5). The concentrations of K + increased in the leachate solutions over the cycles, particularly for C6SL spoils ( Figure 5). An increase in K + production was also observed for the A10RL2, A10RL26, B8RL26, and C4RL26 spoils ( Figure 5). The A10RL2 sample had a higher initial Cl − production (195 mmol L −1 ) among the studied spoils, and the concentrations of Cl − in the leachate solutions of all spoils decreased over the course of experiment ( Figure A4). However, the release of Cl − from C6SL spoils of >2 mm correlated (ρ > 0.90, p < 0.05) with the trajectory of Mg 2+ and total S release ( Figure A5). The dynamics of total S production followed similar patterns to the release of Ca 2+ , particularly for C6SL samples ( Figure A6). The dynamics of Ca 2+ release from the A10RL, B8RL, B9SL, C4RL, and C6SL2 spoils followed a similar pattern to the release of Na + from these spoils (Figures 3 and 4). In contrast, Ca 2+ concentrations in the leachate solutions from the C6SL26 sample increased to 1.4 mmol L −1 at the fourth cycle, before decreasing to 0.08 mmol L −1 at the end of the experiment (Figure 4e). Calcium ion release from the C6SL6 sample followed a similar trend, but the maximum generation of Ca 2+ occurred at the third cycle (Figure 4e).
The release of Mg 2+ correlated (ρ > 0.70, p < 0.05) with the release of Ca 2+ for all spoils, excluding the B8RL26 and C4RL6 spoils ( Figure A3). The trajectory of K + release was spoil specific ( Figure 5). The concentrations of K + increased in the leachate solutions over the cycles, particularly for C6SL spoils ( Figure 5). An increase in K + production was also observed for the A10RL2, A10RL26, B8RL26, and C4RL26 spoils ( Figure 5). The A10RL2 sample had a higher initial Cl − production (195 mmol L −1 ) among the studied spoils, and the concentrations of Cl − in the leachate solutions of all spoils decreased over the course of experiment ( Figure A4). However, the release of Cl − from C6SL spoils of >2 mm correlated (ρ > 0.90, p < 0.05) with the trajectory of Mg 2+ and total S release ( Figure A5). The dynamics of total S production followed similar patterns to the release of Ca 2+ , particularly for C6SL samples ( Figure A6).

Rate of Salt Release from the Coal Mine Spoils
The rate of the release of soluble ions decreased with the increased size of particles for all studied spoils, excluding the B9SL26 sample, where the rate of release of ions was the greatest among the spoils of same type (Table 4). Greater rates of Na + (7.4 mg kg −1 day −1 ) and Cl − (11.54 mg kg −1 day −1 ) release were calculated for the A10RL2 sample ( Table 4). The C6SL2 sample also had higher rates of Mg 2+ , Ca 2+ , and total S release (Table 4).

Rate of Salt Release from the Coal Mine Spoils
The rate of the release of soluble ions decreased with the increased size of particles for all studied spoils, excluding the B9SL26 sample, where the rate of release of ions was the greatest among the spoils of same type (Table 4). Greater rates of Na + (7.4 mg kg −1 day −1 ) and Cl − (11.54 mg kg −1 day −1 ) release were calculated for the A10RL2 sample ( Table 4). The C6SL2 sample also had higher rates of Mg 2+ , Ca 2+ , and total S release (Table 4).

Rate of Salt Release from the Coal Mine Spoils
The rate of the release of soluble ions decreased with the increased size of particles for all studied spoils, excluding the B9SL26 sample, where the rate of release of ions was the greatest among the spoils of same type (Table 4). Greater rates of Na + (7.4 mg kg −1 day −1 ) and Cl − (11.54 mg kg −1 day −1 ) release were calculated for the A10RL2 sample ( Table 4). The C6SL2 sample also had higher rates of Mg 2+ , Ca 2+ , and total S release (Table 4).

Changes in Chemical Properties of the Coal Mine Spoils after Weathering
Firstly, the spoils with similar lithology and geological formation were compared to understand whether different degradation degrees could result in a similar amount of salt reduction in spoils of the same particle size at the end of the experiment. In this study, the amount of salt reduction was spoil specific and size specific (Table 5). However, interestingly, the reductions in the salinity values of all C6SL spoils were comparable (Table 5).
Secondly, the exchangeable cations and soluble ions at the beginning and the end of the experiment were compared to detect the occurrence of adsorption-desorption processes during weathering, as Shaygan et al. [9] reported that this comparison can assist with understanding of the magnitude of adsorption-desorption processes. Remarkably, the final exchangeable Na + and K + for C6SL spoils were greater than the initial values (Table 5). In contrast, their final exchangeable Ca 2+ and Mg 2+ were lower than their initial values (Table 5). Likewise, the final soluble Na + and K + concentrations for C6SL spoils were greater than their initial values (Table 6). Conversely, their final soluble Ca 2+ and Mg 2+ concentrations were lower than their initial values (Table 6).

The Effect of Particle Size on Salt Release from the Coal Mine Spoils
Salts can be dissolved, displaced by other ions, and leached by water percolating through pores [9]. Therefore, the amount, rate, and trajectory of salt release from porous media depend not only on the geochemistry (primary and secondary mineralogy) of the materials but also on the hydrological conditions (hydraulic conductivity, pore size distribution, and connectivity) of the materials; these are influenced by particle size and affect the rate of water flow and, thus, solute mass transport through modifying leaching characteristics (i.e., adsorption-desorption and dissolution rates) [9,31]. Similar to Jurinak et al. [50], who concluded that increasing soil particle size reduces the dissolution rate and salt release, this study showed that spoils with a larger particle size release lower solute masses (Table 4). Hartmann et al. [18] reported that the accessibility of the exchange surface increases with increasing water percolation time in pore systems. In this study, flow rate was inversely correlated with the magnitude and rate of ion release (Figures 1 and 2 and Tables 2-4). Therefore, fast water movement induced by large particles could possibly reduce the contact time for ion exchange between water and spoils' surfaces and decrease the rates of adsorption-desorption and dissolution processes, thus reducing the leaching efficiency and magnitude of the release of solutes from spoils with larger particles.
The breakdown of fragments into smaller particles can change the pore size distribution and connectivity and induce slower hydraulic conductivity [51][52][53][54]. Physical weathering can also break down fragments into clay size particles with larger reactive surfaces [28,54,55]. All the above factors can lead to a higher leaching efficiency of the materials and also lead to greater salt release from porous media. In this study, physical weathering induced by oxidation ( Figure A1) reduced the flow rates of C6SL spoils over the experiment (Figure 1e). Simultaneously, physical weathering increased the amount of clay size particles (Table 3). These factors intensified ion exchange rates and the magnitude of salt release from the C6SL26 and C6SL6 samples (Table 5 and Figures 3 and 4). In this context, Reading et al. [8] reported that prolonged leaching of a pore system can reveal the influence of cation exchange and electrolyte effects on salt release. During the weathering of the C6SL26 and C6SL6 spoils, a reduction in the Ca 2+ and Mg 2+ concentrations of the leachate solutions coincided with an increase in the Na + and K + concentrations of those same solutions, which could indicate the occurrence of ion exchange reactions (Figures 3-5 and A3). Ion exchange was also confirmed by the comparison of exchangeable cations before and after the weathering experiment (Table 5). Therefore, the intensified ion exchange rate, in combination with pyrite oxidation, possibly led to a substantial rise in the leachate salinity for the pyrite-containing samples (C6SL26 and C4SL6) during the fourth and third wet cycles (Figure 2e). These processes also controlled the observed similar reduction in salinity of the C6SL spoils with different particle size classes during the weathering experiment (Table 5). All the above suggest that the initial particle size plays an important role in salt production and release. However, in the case of sulfidic spoils, it is the degree of weathering which mainly governs the magnitude, rate, and dynamics of salt release.

The Effect of Physical and Chemical Weathering on Salt Release from the Coal Mine Spoils
The magnitude of salt release from a porous medium depends on the amount of soluble salts and the dissolution and adsorption-desorption rates [4]. In this study, the A10RL spoils had a higher amount of initial soluble salts (here, NaCl), as they were aged (weathered) spoils collected from the seepage point. These spoils also contained sedimentary sandstone with variable amounts of carbonates and chlorites (Table 1), which could dissolve over time. All the factors mentioned above could elevate the magnitude of salt release from A10RL spoils, particularly for A10RL2 spoils (Table 4 and Figure 2). The same process (dissolution of carbonate compounds) possibly resulted in a cyclic increase of salinity within each cycle for the B8RL26 spoil and caused a sudden rise in the salinity of leachate solutions for the B8RL2 spoil during the sixth cycle ( Figure 2). Therefore, this study found that the original lithology, intrinsic salts, and the presence of authigenic minerals (i.e., carbonates), as well as the magnitude of dissolution and adsorption-desorption processes, can affect the quantity and rate of salt release from coal mine spoils.
The exposure of rocks to atmospheric conditions (wet-dry cycles) can result in the physical weathering of the rocks through slaking layered silicates (i.e., clay minerals) and the expansion of clay minerals, which are associated with moisture absorption, and then dispersion (Price, 1995). The degradation test by Edraki et al. [37] (with 48 h saturation and 28 days drying) showed that the B9SL sample is a dispersive spoil. However, in this study, when the moisture regime was changed (49 mm intensive rainfall and no rain for approximately 30 days), B9SL spoils, which are geologically described as mudrock, showed no evidence of physical weathering and released only small amounts of solutes (Figures 2 and A1). This suggests that the salt generation from coal mine spoils containing mudrock is driven by the moisture regime, which determines the degree of dispersion/physical weathering. In other words, a high water/rock ratio (a high volume of water) or saturation conditions and longer residence times are required for the dispersion of mudrock, which promotes salt production. If the interval of wet cycles was shorter, and/or the duration of leaching (wet cycle) was longer, the magnitude and rate of salt release would increase from mudrock (here, the B8RL and B9SL spoils). Therefore, based on the rock type classification, this study suggests that mudrock spoils generally require a high water/rock ratio for weathering to provide conditions for the production of a considerable amount of solutes.
In some instances, a short exposure of spoils to moisture can result in broad physical weathering [26]. An example of this can be found in reactive, sulfide-bearing rocks in the aerated zone of spoil piles, where the oxidative weathering process can occur and lead to acid generation and the formation of secondary friable minerals (e.g., hydrated sulphates) that exacerbate physical weathering. Given that this chemical weathering process is mineralogy (e.g., pyrite) dependent, it can modify the trajectory of salt release and increase the rate and the magnitude of salt production from coal mine spoils, as was demonstrated in the C6SL26 and C6SL6 spoils ( Figure 2). Exposure of pyrite to moisture, in the presence of oxygen, generated acid ( Figures A2 and A6), which caused the breakdown of spoils to smaller particles ( Figures A1 and A8 and Table 3) and amplified the rate and magnitude of salt release from these spoils ( Figure 2 and Table 4). The onset of salt release depends on the time in which an optimal volume of moisture and oxygen is achieved from the sulfide-bearing spoils. For this reason, the increase in leachate salinity of the C6SL6 sample commenced sooner than those of other C6SL spoils ( Figure 2). The C6SL6 sample had a greater volume of macropores, which trap a greater volume of water, followed by a greater volume of oxygen when water is released, compared with the other C6SL spoils (Table 3). This provided conditions to achieve the optimum volume of oxygen and moisture quicker than other spoils of the same type.
Weathering can create preferred flow pathways within porous media [17]. As water tends to move rapidly through these preferred pathways rather than the soil matrix, at least in wet conditions, a reduction in the release of ions from the pore system occurs [17]. In this study, a large crack which was formed on C6SL2 before the second cycle ( Figure A1) possibly resulted in a rapid reduction of salinity in leachate solutions at the beginning of the second cycle ( Figure 2e). As the spoil material was moistened, the crack was closed by swelling, and flow rates were reduced ( Figure 1e); thus, the salinity of the leachate solutions began to increase (Figure 2e). Although the crack extended throughout the life of the experiment for C6SL2, significant fluctuations were not observed after the second cycle due to the existence of a low amount of soluble salts in the material.

Implications for Management of Coal Mine Spoils
There is continuous discussion about the projection of salt loads into final voids of coal mines and about what the intensity and longevity may be of the impact caused by the salt release from spoil piles containing different minerals that potentially produce saline and/or acidic discharge. This study showed that firstly, the intensity and longevity of salt release to final voids are spoil specific and are related to the original lithology, intrinsic salts, and the presence of authigenic minerals, as well as the magnitude of dissolution and adsorption-desorption processes. For example, the C6SL and C4RL spoils both belonged to the German Creek formation, comprising sandstone and collected from the same mine site. However, the existence of pyrite and its oxidative weathering resulted in a greater release of solutes from the C6SL spoils compared with the C4RL spoils ( Figure 2 and Table 1). Secondly, salt release to final voids is governed by particle size and the hydrological conditions of spoil piles. For instance, the B8RL and B9SL samples, which contained mudrock and were collected from the same mine (Table 1), did not release high amounts of salts due to the moisture-regime-dependent nature of these types of rocks for salt production. This study also confirmed the influence of the hydrological conditions of spoils on salt release through fluctuations in the amount of salt release when a preferred pathway (crack) in the C6SL2 sample occurred.
This study contributes to understanding the mechanisms that control salt generation and release from spoil piles, supporting the use of predictive hydro-geochemical models. The decay values which can be produced based on the results of this study, including the chemistry of leachate solutions (Figures 3-5, A2 and A3), initial geochemistry of the spoils (Table 1), and the time of ion discharge can provide a starting point for the long-term prediction of salt release from spoil piles. Hydro-geochemical models, which can perform and integrate complex scenarios and mechanisms, can be used to predict the intensity and longevity of salt release from a spoil pile to surrounding environments (e.g., final voids). Up until now, modelling of salt generation from mine wastes has mostly been based on simulations assuming homogeneous distributions of hydraulic and geochemical properties [56,57]. However, the spoil piles are heterogeneous with strong spatial and temporal variability in their physical and chemical properties [21] which provide challenges for scaling up and the prediction of salt release from spoil piles. The decay values can be used in a hydro-geochemical model that can also consider a range of particle size fractions and thus assist with more precise estimation of salt production (both spatially and temporally) from a coal mine spoil pile. As an example, the Na + decay values presented in Figure A7 can also be used in hydro-geochemical models for the prediction of salt release. Nevertheless, the decay values require optimisation based on the moisture conditions of the site (i.e., the water/rock ratio), initial hydrological and geochemical characteristics of the spoil pile, and the degree of weathering.
In Australia, similar to other parts of the world, coal mines are required to be rehabilitated either to agricultural or grazing lands, or to be rehabilitated by revegetation, typically using native or local plants to reinstate the disturbed land to a level of ecosystem functionality, preferably to the exact same state as it was prior to coal mining [2,58]. While the types of coal mine spoils are site specific, the typical rehabilitation technique (in Australia) involves creating a landform using spoils, placement of soil (up to a depth of 30 cm [2]) on top of the spoil pile, seeding, and the addition of fertilisers to create conditions for plant growth on the landform [59]. Salt release from coal mine spoils is one of the main factors that can limit plant growth not only by creating saline seepage, but also by upward salt movement during dry periods. This experimental study demonstrated that spoils that had a larger particle size most often generated lower concentrations of saline solutions ( Table 4), suggesting that the placement of a layer of spoils with large particles (>6 mm) in the post-mining landform (below the soil cover) may be an effective strategy in reducing saline seepage and upward movement of salt. During wet periods, the spoils with large particles may release a higher volume of water but fewer solutes, as was discussed previously. During drought conditions, the upward movement of solutes is predicted to be lower within the spoils with large particles, due to the lower potential of creating capillary rise, which is inversely related to pore size and particle size [60][61][62]. This can reduce the intensity and longevity of the impacts potentially caused by salt release. Based on the results of this study, with respect to the moisture regime of the site, the application of nondispersive spoils (spoils with low degradation degree; rock-like spoils) with larger particles may be more beneficial, as they are less impacted by weathering and do not provide suitable conditions for salt release to surrounding environments. Further studies are required to examine the effectiveness of the suggested landform design in relation to the climatic conditions and moisture regime of each specific mine site.

Conclusions
The amount, rate, and dynamics of salt release to the surrounding area of coal mine spoil piles (e.g., final voids and water ways) are defined by the magnitude of geochemical processes such as dissolution and adsorption-desorption, as well as the quantity of soluble salts and the presence of authigenic minerals in the spoils. This study showed that these geochemical processes can be affected by the weathering degree, particle size, and hydrological conditions of the spoils over time, and thus modify the intensity and longevity of salt release to the surrounding environments. If the spoils are not dispersive (i.e., spoils with low degradation degree), the initial particle size governs salt load to the environment, in which salt production is reduced with increasing particle size. In contrast, the degree of weathering controls the salt release from dispersive spoils (i.e., spoils with high degradation degree). In this context, developing a hydro-geochemical model using the calculated decay curves of salt release for known classes of spoils can assist with scaling up and the accurate prediction of salt load to the surrounding areas.