Utilization of Ammonium Chloride as a Novel Selective Depressant in Reverse Flotation of Potassium Chloride

The separation of sylvite (KCl) and halite (NaCl), two main minerals in potash ores, is difficult because of the high ion concentration, fine particles of NaCl, and aggregation of KCl and NaCl in the saturated system. This study employed ammonium chloride (NH4Cl) as a new depressant and dodecyl morpholine as a collector in the reverse flotation process. The depressing mechanisms were studied by adsorption capacity experiments, infrared spectral analysis, and molecular dynamics simulations. The flotation tests showed that NaCl recovery increased to 97% after the addition of NH4Cl, while KCl recovery was reduced to <1%. Notably, NH4Cl not only acted as a selective KCl depressant, but also activated NaCl flotation. The FTIR measurements showed that NH4Cl was physically adsorbed onto the KCl and NaCl surfaces. Adsorption capacity experiments and molecular dynamics simulations confirmed more favorable NH4Cl adsorption on the KCl surface than on the NaCl surface. Moreover, the KCl mineral surface was more hydrophilic, while that of NaCl was more hydrophobic. Relative concentration analysis revealed that >90% ammonium and chloride ions were distributed 2–10 Å away from the KCl surface but were dispersed on the NaCl surface, indicating that NH4Cl exhibited stronger intermolecular interactions with KCl than with NaCl.


Introduction
Potash is a significant raw material in medicine, food, and other chemical industries and is most notably used as a fertilizer in agriculture [1,2]. Among all the mineral salt species used in sylvite (KCl) production, halite (NaCl) separation is very difficult because of its high ion concentration, fine particles, and its aggregation with KCl in a saturated system. Thus, globally, most potash ores are concentrated by froth flotation [3]. In this process, direct flotation collectors comprise alkyl fatty amines [4] and alkyl sulfonates [5], while reverse flotation collectors contain carboxylic acids [6], amides [7], and morpholines [8].
Previous studies [9,10] have shown that in the flotation of KCl, aliphatic primary amines with C12-C18 carbon chain lengths exhibit better flotation and the flotation effect improves as the carbon chain length increases. Currently, over 80% of the world's potash is produced by the selective flotation of KCl from NaCl and other gangue minerals using long-chain (C16-C22 aliphatic chains) amine collectors, as observed in potash mining in Saskatchewan [11]. However, the process of eliminating the C16-C22 long-chain amines is relatively harsh and comprises melting at temperatures in the range 70-90 • C and neutralization with hydrochloric or acetic acids [12]. In the actual flotation process,

Materials and Reagents
Analytically pure KCl and NaCl samples were used as pure minerals. Both samples were dry ground and sieved. The fractions of pure KCl and NaCl with the particle sizes 165-200 µm and <200 µm, respectively, were obtained as flotation samples and used as single minerals. The X-ray powder diffraction spectra revealed that the purities of the KCl and NaCl samples were 98% and 99%, respectively. Both minerals were ground in a ceramic ball mill and the fractions in the size range 38-74 µm were used in the flotation tests. Deionized water with a resistivity >18 MΩ·cm was used in all the experiments. The reagents used in the study are listed in Table 1. The saturated brines were prepared by dissolving a sufficient amount of salt (Table 2) in Millipore water with stirring for 2 h. The saturated solutions were then stored at room temperature for 24 h and filtered before use. The DMP solution (0.01 g/L, 99% purity) was prepared in the laboratory with Minerals 2019, 9,41 3 of 12 saturated brine as the solvent. Flotation tests were conducted using DMP as the collector and NH 4 Cl as the regulator. All the reagents used in the tests were of analytical grade.

Flotation Tests
The single-mineral flotation KCl and NaCl tests were conducted on an XFG flotation machine (Exploring Machinery Plant, Changchun, China) with a volume of 50 mL and an impeller speed of 1850 rpm (Figure 1). A well-configured and filtered saturated brine solution was used as the flotation medium. In each test, 10 g mineral sample and 40 mL saturated brine solution were added in the cell and conditioned for 3 min. The flotation experiment was divided into two parts: first, the suspension was dosed with 900-1500 g/t DMP as the collector. Subsequently, to test the effect of NH 4 Cl, the prepared NH 4 Cl was applied before adding the optimal DMP dosage. The depressant (if needed) and collector were added sequentially and conditioned for 5 and 10 min, respectively. After 10 min of flotation, the products were collected, dried, and weighed and the recovery was then calculated. The flotation flowsheet is displayed in Figure 2. Each flotation test was conducted in triplicate and the average was reported as the final value. The standard deviation, which is presented as an error bar, was obtained from the mean of the three measurements per experimental condition.

Flotation Tests
The single-mineral flotation KCl and NaCl tests were conducted on an XFG flotation machine (Exploring Machinery Plant, Changchun, China) with a volume of 50 mL and an impeller speed of 1850 rpm (Figure 1). A well-configured and filtered saturated brine solution was used as the flotation medium. In each test, 10 g mineral sample and 40 mL saturated brine solution were added in the cell and conditioned for 3 min. The flotation experiment was divided into two parts: first, the suspension was dosed with 900-1500 g/t DMP as the collector. Subsequently, to test the effect of NH4Cl, the prepared NH4Cl was applied before adding the optimal DMP dosage. The depressant (if needed) and collector were added sequentially and conditioned for 5 and 10 min, respectively. After 10 min of flotation, the products were collected, dried, and weighed and the recovery was then calculated. The flotation flowsheet is displayed in Figure 2. Each flotation test was conducted in triplicate and the average was reported as the final value. The standard deviation, which is presented as an error bar, was obtained from the mean of the three measurements per experimental condition.

Adsorption Measurements
The adsorption measurements were completed using a total organic carbon analyzer (TOC-L CPH/CPN, Shimadzu Corporation, Kyoto, Japan). Different dosages were used as the dosing conditions, while a saturated salt brine solution with no added minerals was set as the reference group. A total of 10 g mineral was added to the experimental group, while no minerals were added to the reference group. The volume of the entire slurry was maintained at 40 mL after the addition of the mineral (in different dosages). All the pulp was placed in a clean beaker and stirred in a magnetic stirrer for 15 min, after which the supernatant was withdrawn. The sample was diluted four-fold in a 50 mL PET tube and centrifuged for 15 min at a speed of 18,000 rpm to determine the total amount of organic carbon. The experiments were repeated at least in triplicate and the average data were plotted. The standard deviation, presented as an error bar, was obtained using the mean of the three measurements per experimental condition. The adsorption of the reagent on the mineral surface was calculated as follows [16,17]: where Γ is the adsorption amount, μg/g mineral; m1 is the weight of total reagent in the solution measured in the reference group, μg; m2 is the weight of total reagent in the solution measured in the experimental group, μg; and m is the weight of the mineral, g.

Infrared Spectral Analysis
Fourier-transform infrared spectroscopy (FTIR, Bio-Rad FTS-6000, Cambridge, MA, USA) was performed at room temperature in the range 4000-400 cm −1 . After the pure minerals were conditioned with the reagents in the solution, the samples were filtered naturally, dried thoroughly in a vacuum oven at 40 °C, and ground to ˂2 μm in an agate mortar. The spectra of the solids were recorded using KBr pellets.

MD Simulations
The adsorption behavior of the ammonium ions on the NaCl (100) and KCl (001) surfaces was explored by MD simulation. The initial adsorption models were built using the software package, Materials Studio 6.0 (Dassault Systèmes BIOVIA, San Diego, CA, USA). The condensed-phase optimized molecular potentials for atomic simulation studies (COMPASS) force-field was used to calculate the inter-and intra-atomic interactions. Additionally, density functional theory calculations for the ammonium ion, NaCl (100) surface, and KCl (001) surface were implemented to assign the charges in the CASTEP module. The Perdew-Burke-Ernzerhof functional with a generalized gradient approximation was used throughout the study.
Based on the natural cleavage plane reported in previous studies [18,19], the NaCl and KCl mineral surfaces were built to construct the adsorption models. After the mineral surface models

Adsorption Measurements
The adsorption measurements were completed using a total organic carbon analyzer (TOC-L CPH/CPN, Shimadzu Corporation, Kyoto, Japan). Different dosages were used as the dosing conditions, while a saturated salt brine solution with no added minerals was set as the reference group. A total of 10 g mineral was added to the experimental group, while no minerals were added to the reference group. The volume of the entire slurry was maintained at 40 mL after the addition of the mineral (in different dosages). All the pulp was placed in a clean beaker and stirred in a magnetic stirrer for 15 min, after which the supernatant was withdrawn. The sample was diluted four-fold in a 50 mL PET tube and centrifuged for 15 min at a speed of 18,000 rpm to determine the total amount of organic carbon. The experiments were repeated at least in triplicate and the average data were plotted. The standard deviation, presented as an error bar, was obtained using the mean of the three measurements per experimental condition. The adsorption of the reagent on the mineral surface was calculated as follows [16,17]: where Γ is the adsorption amount, µg/g mineral; m 1 is the weight of total reagent in the solution measured in the reference group, µg; m 2 is the weight of total reagent in the solution measured in the experimental group, µg; and m is the weight of the mineral, g.

Infrared Spectral Analysis
Fourier-transform infrared spectroscopy (FTIR, Bio-Rad FTS-6000, Cambridge, MA, USA) was performed at room temperature in the range 4000-400 cm −1 . After the pure minerals were conditioned with the reagents in the solution, the samples were filtered naturally, dried thoroughly in a vacuum oven at 40 • C, and ground to <2 µm in an agate mortar. The spectra of the solids were recorded using KBr pellets.

MD Simulations
The adsorption behavior of the ammonium ions on the NaCl (100) and KCl (001) surfaces was explored by MD simulation. The initial adsorption models were built using the software package, Materials Studio 6.0 (Dassault Systèmes BIOVIA, San Diego, CA, USA). The condensed-phase optimized molecular potentials for atomic simulation studies (COMPASS) force-field was used to calculate the inter-and intra-atomic interactions. Additionally, density functional theory calculations for the ammonium ion, NaCl (100) surface, and KCl (001) surface were implemented to assign the charges in the CASTEP module. The Perdew-Burke-Ernzerhof functional with a generalized gradient approximation was used throughout the study.
Based on the natural cleavage plane reported in previous studies [18,19], the NaCl and KCl mineral surfaces were built to construct the adsorption models. After the mineral surface models were built, the amorphous cell, a packing module, was used for the construction of the reagent matrix on the surfaces. The chloride ions were added to maintain the system neutral. A strict reagent addition sequence ( Figure 1) was adopted. The built vacuum slab (thickness ca. 80 Å) eliminated the influence of the period boundary conditions. During the simulation, the ammonium and chloride ions in the solution moved freely, while the NaCl and KCl sheets were fixed.
Molecular optimization was carried out using a Smart Minimizer to eliminate the possible overlap of molecules during the process of building the configuration. The simple point-charge model was used to simulate liquid water. The temperature was controlled by an Andersen thermostat. The canonical (NVT) ensemble was applied at the temperature 298 K for each system using a Nosé-Hoover thermostat, while the integration step was set at 1 fs. Finally, each system was simulated for 4 ns to reach the equilibrium state and the last 1 ns of each trajectory was used for the analysis. Figure 3a illustrates the effect of the DMP dosage on the flotation of KCl and NaCl in the absence of NH 4 Cl. In this flotation system, the NaCl exhibited good flotability, while KCl presented a very poor flotation ability in the presence of the DMP collector. Further, with an increase in the collector dosage, in the range 900-1400 g/t, the flotation recovery increased monotonically and a maximum value of 93.13% NaCl recovery was achieved at 1400 g/t DMP. In the test dosage range 900-1500 g/t, the flotation recoveries of NaCl and KCl were >70% and <4.5%, respectively. Comparatively, typical recoveries achieved in some of the flotation plants in Saskatchewan, which used long-chain amines as collectors, were in the range 85-88% [11]. Qinghai Avic Resources Co., Ltd., in Mahai also employed traditional KCl flotation from NaCl with an amine as the collector to produce the final KCl product in the grade range 86-92% [13]. In this experiment, the loss of KCl was >10% when long-chain amines were used as collectors, exceeding the maximum value (4.5%) of KCl loss rate. This indicates that DMP exhibits a good selective collecting ability for NaCl and can be utilized to separate NaCl from KCl. were built, the amorphous cell, a packing module, was used for the construction of the reagent matrix on the surfaces. The chloride ions were added to maintain the system neutral. A strict reagent addition sequence (Figure 1) was adopted. The built vacuum slab (thickness ca. 80 Å ) eliminated the influence of the period boundary conditions. During the simulation, the ammonium and chloride ions in the solution moved freely, while the NaCl and KCl sheets were fixed. Molecular optimization was carried out using a Smart Minimizer to eliminate the possible overlap of molecules during the process of building the configuration. The simple point-charge model was used to simulate liquid water. The temperature was controlled by an Andersen thermostat. The canonical (NVT) ensemble was applied at the temperature 298 K for each system using a Nosé-Hoover thermostat, while the integration step was set at 1 fs. Finally, each system was simulated for 4 ns to reach the equilibrium state and the last 1 ns of each trajectory was used for the analysis. Figure 3a illustrates the effect of the DMP dosage on the flotation of KCl and NaCl in the absence of NH4Cl. In this flotation system, the NaCl exhibited good flotability, while KCl presented a very poor flotation ability in the presence of the DMP collector. Further, with an increase in the collector dosage, in the range 900-1400 g/t, the flotation recovery increased monotonically and a maximum value of 93.13% NaCl recovery was achieved at 1400 g/t DMP. In the test dosage range 900-1500 g/t, the flotation recoveries of NaCl and KCl were >70% and ˂4.5%, respectively. Comparatively, typical recoveries achieved in some of the flotation plants in Saskatchewan, which used long-chain amines as collectors, were in the range 85%-88% [11]. Qinghai Avic Resources Co., Ltd., in Mahai also employed traditional KCl flotation from NaCl with an amine as the collector to produce the final KCl product in the grade range 86%-92% [13]. In this experiment, the loss of KCl was >10% when longchain amines were used as collectors, exceeding the maximum value (4.5%) of KCl loss rate. This indicates that DMP exhibits a good selective collecting ability for NaCl and can be utilized to separate NaCl from KCl.  The flotation response was significantly different when NH4Cl was employed as the depressant. Figure 3b illustrates the effect of NH4Cl dosage on the flotation of KCl and NaCl in the presence of DMP. The DMP dosage was fixed at 1300 g/t, the optimal dosage in the DMP flotation system. The KCl and NaCl recoveries were 4.23% and 89.88%, respectively, in the absence of NH4Cl. After the addition of NH4Cl, the flotation of KCl was inhibited, while the recovery was ˂1% in the test dosage  The flotation response was significantly different when NH 4 Cl was employed as the depressant. Figure 3b illustrates the effect of NH 4 Cl dosage on the flotation of KCl and NaCl in the presence of DMP. The DMP dosage was fixed at 1300 g/t, the optimal dosage in the DMP flotation system. The KCl and NaCl recoveries were 4.23% and 89.88%, respectively, in the absence of NH 4 Cl. After the addition of NH 4 Cl, the flotation of KCl was inhibited, while the recovery was <1% in the test dosage range 500-2500 g/t. Interestingly, NH 4 Cl did not depress the flotation of NaCl but showed an activation Minerals 2019, 9, 41 6 of 12 effect on NaCl. The recovery of NaCl increased with an increase in the NH 4 Cl dosage and reached a maximum value of 97.55% at an NH 4 Cl dosage of 2000 g/t, indicating that NH 4 Cl promoted the flotation separation of NaCl and KCl with DMP as a collector. Figure 4a presents the adsorption capacity of DMP on the NaCl and KCl surfaces. DMP adsorption was significantly higher onto the NaCl surface than on the KCl surface. The adsorption of DMP onto KCl and NaCl gradually increased and reached maximum values at dosages of 1300 and 1400 g/t, respectively. These values agree with the flotation experiment results. Figure 4b illustrates that in the dosage range tested (500-2500 g/t), the adsorption capacity of NH 4 Cl onto the KCl surface was significantly higher than onto the NaCl surface. The adsorption amount of NH 4 Cl onto the KCl surface was about 50 µg/g higher than that on the NaCl surface and increased with an increase in the NH 4 Cl dosage. The difference between the adsorption amounts of the two minerals reached a maximum of 71.9 µg/g at the NH 4 Cl dosage 2500 g/t.

Adsorption Capacity Experiments
Minerals 2019, 9, x FOR PEER REVIEW 6 of 12 range 500-2500 g/t. Interestingly, NH4Cl did not depress the flotation of NaCl but showed an activation effect on NaCl. The recovery of NaCl increased with an increase in the NH4Cl dosage and reached a maximum value of 97.55% at an NH4Cl dosage of 2000 g/t, indicating that NH4Cl promoted the flotation separation of NaCl and KCl with DMP as a collector. Figure 4a presents the adsorption capacity of DMP on the NaCl and KCl surfaces. DMP adsorption was significantly higher onto the NaCl surface than on the KCl surface. The adsorption of DMP onto KCl and NaCl gradually increased and reached maximum values at dosages of 1300 and 1400 g/t, respectively. These values agree with the flotation experiment results. Figure 4b illustrates that in the dosage range tested (500-2500 g/t), the adsorption capacity of NH4Cl onto the KCl surface was significantly higher than onto the NaCl surface. The adsorption amount of NH4Cl onto the KCl surface was about 50 μg/g higher than that on the NaCl surface and increased with an increase in the NH4Cl dosage. The difference between the adsorption amounts of the two minerals reached a maximum of 71.9 μg/g at the NH4Cl dosage 2500 g/t.

FTIR Spectroscopy Analysis
FTIR is commonly used to characterize the mechanism of the reaction between a mineral and a reagent. The FTIR spectra of NH4Cl, KCl, and KCl reacted with NH4Cl are presented in Figure 5. In the NH4Cl spectrum, the characteristic sharp band near 3137 cm −1 corresponds to the stretching N-H vibration, while the band near 1402 cm −1 was assigned to the flexural vibration of the ammonium ion [20]. After the reaction between KCl and NH4Cl (Figure 5c

FTIR Spectroscopy Analysis
FTIR is commonly used to characterize the mechanism of the reaction between a mineral and a reagent. The FTIR spectra of NH 4 Cl, KCl, and KCl reacted with NH 4 Cl are presented in Figure 5. In the NH 4 Cl spectrum, the characteristic sharp band near 3137 cm −1 corresponds to the stretching N-H vibration, while the band near 1402 cm −1 was assigned to the flexural vibration of the ammonium ion [20]. After the reaction between KCl and NH 4 Cl (Figure 5c The DMP FTIR spectra are presented in Figure 6a. The bands at 2922 and 2852 cm −1 were assigned to the C-H stretching vibration [21,22]. Moreover, weak characteristic peaks of DMP at 3026 and 2802 cm −1 were detected on the KCl surface under the precondition that NH4Cl reacted with KCl, suggesting that small amounts of DMP adsorbed onto the KCl surface. The characteristic peaks of DMP at 3029 and 2802 cm −1 were still significant in the NaCl+NH4Cl spectrum (Figure 6e), indicating that DMP was effectively adsorbed onto the NaCl surface in the presence of NH4Cl. Based on these results, it can be concluded that significantly strong physical adsorption of DMP occurs on the NaCl surface after treatment with NH4Cl. On the other hand, DMP hardly adsorbs on the NH4Cl-treated KCl surface. These FTIR results support the flotation results.

NH4Cl Adsorption States on the NaCl and KCl Crystal Surfaces in Vacuum
MD simulation was used to investigate the solid-liquid interface properties and determine the micro-adsorption structure on a molecular scale [23][24][25][26]. Studies have reported that NH4Cl adsorbs  The DMP FTIR spectra are presented in Figure 6a. The bands at 2922 and 2852 cm −1 were assigned to the C-H stretching vibration [21,22]. Moreover, weak characteristic peaks of DMP at 3026 and 2802 cm −1 were detected on the KCl surface under the precondition that NH 4 Cl reacted with KCl, suggesting that small amounts of DMP adsorbed onto the KCl surface. The characteristic peaks of DMP at 3029 and 2802 cm −1 were still significant in the NaCl+NH 4 Cl spectrum (Figure 6e), indicating that DMP was effectively adsorbed onto the NaCl surface in the presence of NH 4 Cl. Based on these results, it can be concluded that significantly strong physical adsorption of DMP occurs on the NaCl surface after treatment with NH 4 Cl. On the other hand, DMP hardly adsorbs on the NH 4 Cl-treated KCl surface. These FTIR results support the flotation results. The DMP FTIR spectra are presented in Figure 6a. The bands at 2922 and 2852 cm −1 were assigned to the C-H stretching vibration [21,22]. Moreover, weak characteristic peaks of DMP at 3026 and 2802 cm −1 were detected on the KCl surface under the precondition that NH4Cl reacted with KCl, suggesting that small amounts of DMP adsorbed onto the KCl surface. The characteristic peaks of DMP at 3029 and 2802 cm −1 were still significant in the NaCl+NH4Cl spectrum (Figure 6e), indicating that DMP was effectively adsorbed onto the NaCl surface in the presence of NH4Cl. Based on these results, it can be concluded that significantly strong physical adsorption of DMP occurs on the NaCl surface after treatment with NH4Cl. On the other hand, DMP hardly adsorbs on the NH4Cl-treated KCl surface. These FTIR results support the flotation results.

NH4Cl Adsorption States on the NaCl and KCl Crystal Surfaces in Vacuum
MD simulation was used to investigate the solid-liquid interface properties and determine the micro-adsorption structure on a molecular scale [23][24][25][26]. Studies have reported that NH4Cl adsorbs

NH 4 Cl Adsorption States on the NaCl and KCl Crystal Surfaces in Vacuum
MD simulation was used to investigate the solid-liquid interface properties and determine the micro-adsorption structure on a molecular scale [23][24][25][26]. Studies have reported that NH 4 Cl adsorbs on KCl through the chloride ions, which act with the surface potassium atoms. Figure 7 presents the models of the mineral-reagent complex after NH 4 Cl adsorption in vacuum. Almost all the chloride ions absorbed onto the KCl surface and interacted strongly with the potassium atoms. Moreover, the ammonium ions were also close to the KCl surface, probably because of their interaction with the surface chloride ions and the chlorine atoms in the KCl crystal. On the other hand, in the NaCl system (Figure 7b), the ammonium and chloride ions were alienated from the surface and did not collect at the NaCl surface. on KCl through the chloride ions, which act with the surface potassium atoms. Figure 7 presents the models of the mineral-reagent complex after NH4Cl adsorption in vacuum. Almost all the chloride ions absorbed onto the KCl surface and interacted strongly with the potassium atoms. Moreover, the ammonium ions were also close to the KCl surface, probably because of their interaction with the surface chloride ions and the chlorine atoms in the KCl crystal. On the other hand, in the NaCl system (Figure 7b), the ammonium and chloride ions were alienated from the surface and did not collect at the NaCl surface. The color scheme is as follows: red, oxygen atoms; white, hydrogen atoms; blue, nitrogen atoms; purple, potassium atoms; yellow, sodium atoms; and green, chlorine atoms.

NH4Cl Adsorption States on the NaCl and KCl Crystal Surfaces in the Presence of Water Molecules
Retaining the original vacuum condition, 200 water molecules were added to the system to investigate the effects of the ammonium and chloride ions on the KCl and NaCl surfaces in the presence of water as well as the hydrophilicity and hydrophobicity of the two minerals. Figure 8a illustrates that the ammonium and chloride ions interacted strongly with KCl and were effectively adsorbed onto the mineral surface. On the other hand, in the KCl system, the water molecules were close to the mineral surface because KCl is hydrophilic. The data in Figure 8b revealed that in the presence of water molecules, the ammonium and chloride ions were further away from the NaCl surface because of the hydrophobicity of NaCl. The above conditions were consistent with those observed under vacuum conditions and validated the conclusion reached on flotation.

NH 4 Cl Adsorption States on the NaCl and KCl Crystal Surfaces in the Presence of Water Molecules
Retaining the original vacuum condition, 200 water molecules were added to the system to investigate the effects of the ammonium and chloride ions on the KCl and NaCl surfaces in the presence of water as well as the hydrophilicity and hydrophobicity of the two minerals. Figure 8a illustrates that the ammonium and chloride ions interacted strongly with KCl and were effectively adsorbed onto the mineral surface. On the other hand, in the KCl system, the water molecules were close to the mineral surface because KCl is hydrophilic. The data in Figure 8b revealed that in the presence of water molecules, the ammonium and chloride ions were further away from the NaCl surface because of the hydrophobicity of NaCl. The above conditions were consistent with those observed under vacuum conditions and validated the conclusion reached on flotation. The analysis of the relative concentrations of the reagents as a function of distance from the mineral surface provided a quantitative basis for the adsorption ability [27]. The NH4Cl distribution on different minerals could therefore be quantified by calculating the relative concentrations of the ammonium and chloride ions [28]. The concentration profiles along the z-axis were calculated with the normal z-axis on the mineral surface set as the zero point ( Figure 9). After the interaction of NH4Cl with KCl, the ammonium and chloride ions were both distributed at distances in the range 2-13 Å from the KCl surface and 2-19 Å from the NaCl surface. The relative concentrations of the ammonium ions in the range 3-8 Å away from the KCl surface are all >10%, while the maximum relative concentration of ammonium ions 9.46 Å away from the NaCl surface is ˂10%. Moreover, >90% ammonium ions and chloride ions were distributed 2-10 Å away from the KCl surface. However, these ions were dispersed on the NaCl surface, indicating that NH4Cl exhibited a stronger intermolecular interaction with KCl than with NaCl. The relative concentration distribution of the chloride ions on the KCl and NaCl surfaces was similar to that of the ammonium ions. In the aqueous solution system, analysis of the relative concentration results revealed that the ammonium and chloride ions were close to the KCl surface, indicating that the strong interaction between NH4Cl and KCl led to the strong adsorption of the former mineral onto the mineral surface. Conversely, the ammonium and chloride ions were far from the NaCl surface, indicating that NH4Cl exhibited a weaker intermolecular interaction with NaCl than with KCl. The results of the MD simulations verified the flotation results. The analysis of the relative concentrations of the reagents as a function of distance from the mineral surface provided a quantitative basis for the adsorption ability [27]. The NH 4 Cl distribution on different minerals could therefore be quantified by calculating the relative concentrations of the ammonium and chloride ions [28]. The concentration profiles along the z-axis were calculated with the normal z-axis on the mineral surface set as the zero point ( Figure 9). After the interaction of NH 4 Cl with KCl, the ammonium and chloride ions were both distributed at distances in the range 2-13 Å from the KCl surface and 2-19 Å from the NaCl surface. The relative concentrations of the ammonium ions in the range 3-8 Å away from the KCl surface are all >10%, while the maximum relative concentration of ammonium ions 9.46 Å away from the NaCl surface is <10%. Moreover, >90% ammonium ions and chloride ions were distributed 2-10 Å away from the KCl surface. However, these ions were dispersed on the NaCl surface, indicating that NH 4 Cl exhibited a stronger intermolecular interaction with KCl than with NaCl. The relative concentration distribution of the chloride ions on the KCl and NaCl surfaces was similar to that of the ammonium ions. In the aqueous solution system, analysis of the relative concentration results revealed that the ammonium and chloride ions were close to the KCl surface, indicating that the strong interaction between NH 4 Cl and KCl led to the strong adsorption of the former mineral onto the mineral surface. Conversely, the ammonium and chloride ions were far from the NaCl surface, indicating that NH 4 Cl exhibited a weaker intermolecular interaction with NaCl than with KCl. The results of the MD simulations verified the flotation results. The color scheme is as follows: red, oxygen atoms; white, hydrogen atoms; blue, nitrogen atoms; purple, potassium atoms; yellow, sodium atoms; and green, chlorine atoms.
The analysis of the relative concentrations of the reagents as a function of distance from the mineral surface provided a quantitative basis for the adsorption ability [27]. The NH4Cl distribution on different minerals could therefore be quantified by calculating the relative concentrations of the ammonium and chloride ions [28]. The concentration profiles along the z-axis were calculated with the normal z-axis on the mineral surface set as the zero point ( Figure 9). After the interaction of NH4Cl with KCl, the ammonium and chloride ions were both distributed at distances in the range 2-13 Å from the KCl surface and 2-19 Å from the NaCl surface. The relative concentrations of the ammonium ions in the range 3-8 Å away from the KCl surface are all >10%, while the maximum relative concentration of ammonium ions 9.46 Å away from the NaCl surface is ˂10%. Moreover, >90% ammonium ions and chloride ions were distributed 2-10 Å away from the KCl surface. However, these ions were dispersed on the NaCl surface, indicating that NH4Cl exhibited a stronger intermolecular interaction with KCl than with NaCl. The relative concentration distribution of the chloride ions on the KCl and NaCl surfaces was similar to that of the ammonium ions. In the aqueous solution system, analysis of the relative concentration results revealed that the ammonium and chloride ions were close to the KCl surface, indicating that the strong interaction between NH4Cl and KCl led to the strong adsorption of the former mineral onto the mineral surface. Conversely, the ammonium and chloride ions were far from the NaCl surface, indicating that NH4Cl exhibited a weaker intermolecular interaction with NaCl than with KCl. The results of the MD simulations verified the flotation results. The simulation results in this study were consistent with the surface charge theory [29], where the governing mechanism for soluble salt flotation comprised the electrostatic interaction between the salt surface and the collector species [30]. In the surface charge theory, KCl and NaCl carry opposite charges, KCl being negatively charged and NaCl being positively charged in their saturated brines. These speculations were later confirmed by the nonequilibrium electrophoretic mobility measurements [31]. Therefore, the flotation separation of KCl from NaCl was achieved because of the adsorption of cationic surfactants such as C12-C18 amine ions or positively charged collector colloids on the negatively charged KCl surface but not on the positively charged NaCl surface [31,32]. MD simulation results revealed that the adsorption of chloride ions onto the KCl surface resulted in a negatively charged KCl surface. The results verified the above theory that the KCl surface was negatively charged; thus, ammonium ions were adsorbed onto the negatively charged KCl surface by electrostatic attraction and alienated the positively charged NaCl surface because of electrostatic repulsion.

Conclusions
Dodecyl morpholine is a selective collector for the flotation of NaCl and KCl. Reverse flotation is being applied in the industry and is more effective than traditional flotation, which employs long-chain ammonium compounds as collectors [13,33]. The addition of NH 4 Cl promoted the separation of KCl and NaCl. The flotation tests revealed that NH 4 Cl exerts a considerable inhibitory effect on KCl so that the recovery was reduced to <1%. It also exhibited a significant activation effect on NaCl so that the maximum recovery value was 97.55%.
The adsorption experiments and FTIR measurements indicated that DMP adsorption was significantly higher on the NaCl surface than on the KCl surface. However, the adsorption of NH 4 Cl onto the KCl surface was stronger than that on the NaCl surface and the difference increased with an increase in the NH 4 Cl dosage. The characteristic DMP peaks at~3029 cm −1 and~2802 cm −1 were still significant on the NaCl surface. On the other hand, the characteristic peaks detected on the KCl surface in the presence of NH 4 Cl were weak. Notably, NH 4 Cl and DMP were physically adsorbed onto the surfaces of both minerals.
MD simulations demonstrated that the adsorption of NH 4 Cl onto the KCl surface was very strong because of the presence of electrostatic forces. However, NH 4 Cl displayed weak interactions with NaCl and could not be adsorbed onto the NaCl surface. In an aqueous solution system, the KCl surface was hydrophilic, while the NaCl surface was hydrophobic. The relative concentration measurements indicated that the ammonium and chloride ions were close to the KCl surface but far away from that of NaCl, indicating that NH 4 Cl exhibited stronger interactions with KCl than with NaCl; this led to its stronger adsorption onto the KCl surface. The results in this study therefore confirmed that NH 4 Cl promotes the flotation separation of NaCl and KCl with DMP as the collector.