Synthesis, Structure and NH 3 Sorption Properties of Mixed Mg 1 − x Mn x (NH 3 ) 6 Cl 2 Ammines

: This paper describes the synthesis, crystal structure, and NH 3 sorption properties of Mg 1 − x Mn x (NH 3 ) 6 Cl 2 ( x = 0–1) mixed metal halide ammines, with reversible NH 3 storage capacity in the temperature range 20–350 ◦ C. The stoichiometry ( x ) dependent NH 3 desorption temperatures were monitored using in situ synchrotron radiation powder X-ray di ﬀ raction, thermogravimetric analysis, and di ﬀ erential scanning calorimetry. The thermal analyses reveal that the NH 3 release temperatures decrease in the mixed metal halide ammines in comparison to pure Mg(NH 3 ) 6 Cl 2 , approaching the values of Mn(NH 3 ) 6 Cl 2 . Desorption occurs in three steps of four, one and one NH 3 moles, with the corresponding activation energies of 54.8 kJ · mol − 1 , 73.2 kJ · mol − 1 and 91.0 kJ · mol − 1 in Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 , which is signiﬁcantly lower than the NH 3 release activation energies of Mg(NH 3 ) 6 Cl 2 (E a = 60.8 kJ · mol − 1 , 74.8 kJ · mol − 1 and 91.8 kJ · mol − 1 ). This work shows that Mg 1 − x Mn x (NH 3 ) y Cl 2 ( x = 0 to 1, y = 0 to 6) is stable within the investigated temperature range (20–350 ◦ C) and also upon NH 3 cycling.


Introduction
Energy storage materials and methods have gained high interest to ensure the transition to carbon-free future. Hydrogen, as a high-density energy carrier alternative to fossil fuels, is one of the promising solutions for energy storage systems via solid storage of hydrogen [1][2][3][4]. Several studies have highlighted the potential of ammonia for hydrogen-based energy systems [5][6][7][8].
The ammonia release temperatures are associated with the binding energy of NH 3 with its surrounding ions, which depends on the elements and crystal structures of the metal halides as elucidated in a recent study [24]. Formation of solid solutions has been suggested as an approach to tailor the NH 3 desorption temperatures and kinetics [25][26][27]. The NH 3 binding energies were investigated in SrCl 2 -CaCl 2 solid solutions, and they were found to be intermediate those of the two precursors [28]. This led to studies on solid solutions of Sr 1−x Ba x (NH 3 ) 8 Cl 2 and Sr 1−x Ca x (NH 3 ) 8 Cl 2 , and their respective NH 3 release properties. Sr 1−x Ba x (NH 3 ) 8 Cl 2 solid solutions showed that varying the relative ratios of metal allowed tuning of the desorption temperature of ammonia. The gradual effect on the ammonia release temperature was observed with the optimal mixing condition of 37.5 % of BaCl 2 in Sr 1−x Ba x (NH 3 ) 8 Cl 2 showing the full release of ammonia at temperature T < 100 • C of the final mixed metal halide ammine [25]. Similarly, it was demonstrated for Sr 1−x Ca x Cl 2 solid solutions and the corresponding Sr 1−x Ca x (NH 3 ) 8 Cl 2 ammines that the NH 3 absorption and desorption properties could be enhanced by tuning the mixing ratio [26,27]. Additionally, the ammonia storage properties and crystal structures of the CaCl 2 -CaBr 2 , SrCl 2 -SrBr 2 and SrCl 2 -SrI 2 solid solutions have also been investigated, and intermediate ammonia storage properties of the mixed anion metal halides were observed [28][29][30]. These studies show the possibility of forming mixed metal halides with tunable ammonia sorption properties. Solid solutions of borohydride-based ammines have also been investigated as potential solid-state hydrogen storage materials. The solid solutions of Mg 1−x Mn x (BH 4 ) 2 ·6NH 3 and structural similarities of Mg(BH 4 ) 2 and Mn(BH 4 ) 2 and their corresponding ammines were studied [31,32]. Similar to the present study it revealed temperature changes for ammonia release when compared to those of the pristine samples.
Inspired by the structural similarities between Mg(NH 3 ) 6 Cl 2 and Mn(NH 3 ) 6 Cl 2 , this work addresses an investigation of solid solutions of Mg 1−x Mn x (NH 3 ) 6 Cl 2 . Here we show the synthesis of these novel series of mixed metal halide ammines with tunable properties for the NH 3 desorption. We present the Mg 1−x Mn x (NH 3 ) 6 Cl 2 (x = 0.025, 0.05, 0.1, 0.3 and 0.5) solid solutions obtained by mechanical mixing of MgCl 2 and MnCl 2 , followed by annealing and subsequent exposure to anhydrous NH 3 gas. The mixed metal halide ammines were systematically investigated with in situ powder X-ray diffraction, thermogravimetric analysis, differential scanning calorimetry and volumetric Sieverts techniques. The thermally induced ammonia release for the mixed metal halide ammines is discussed: three NH 3 desorption events are observed and the crystal structures of the intermediate ammine phases are identified and structurally characterized. The kinetics, absorption, and desorption properties of NH 3 are studied. The results presented in this work show that by changing the relative Mg/Mn ratio the NH 3 sorption properties can be tuned and optimized depending on the application.

Sample Preparation
Anhydrous MgCl 2 and MnCl 2 powders with a purity of 99.999% were purchased from Alfa Aesar and Sigma-Aldrich, respectively. Mg 1−x Mn x Cl 2 solid solutions (x = 0.025, 0.05, 0.1, 0.3 and 0.5) were obtained using a SPEX SamplePrep 8000D Dual Mixer high-energy ball mill. The powders were placed in a 25 mL hardened steel vial together with hardened steel balls (10 mm diameter) in a ball-to-powder mass ratio of 16:1 and sealed in an Ar-filled glove box (<1 ppm of O 2 and H 2 O). The ball milling program was for one hour.
The as-milled powders were annealed to increase the crystallinity. Batches of~0.5 g of the as-milled powders were sealed in a stainless-steel cylinder inside a glove box, and subsequently heated to 350 • C with a heating rate of 1 • C·sec −1 and kept isothermal at 350 • C for 24 h. These samples are denoted "as-synthesized" samples. Subsequently, the as-synthesized samples were placed in a high temperature stainless-steel cylinder and connected to an in-house built Sieverts apparatus. The samples were then exposed to an NH 3 gas pressure of 2.5 bar at room temperature (RT) for at least 3 h. MnCl 2 was ammoniated for 3 h at T = −20 • C and 1 bar NH 3 . It was then stored in a glovebox freezer at −34 • C prior to the experiments, due to instability of the Mn(NH 3 ) 6 Cl 2 at ambient conditions. These samples are denoted "ammoniated" samples.

Thermal Analysis
Combined thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) of the monometallic and mixed metal halide ammines were measured using a Netzsch STA 449 F3 Jupiter apparatus. The powders (~40 mg) were placed in an alumina crucible with a pierced lid under protective Ar atmosphere in a glove box. The alumina crucibles were shortly exposed to air (ca. 1 min) during mounting in the TGA-DSC apparatus. The powders were heated from RT to 455 • C with a heating rate of 5 • C·min −1 in an Ar flow of 50 mL·min −1 . Additionally, the batches of Mg(NH 3 ) 6 Cl 2 , Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 and Mn(NH 3 ) 6 Cl 2 powder (~10 mg) were measured at six different heating rates of 1, 2, 5, 10, 20 and 40 • C·min −1 for Kissinger analyses [33].

Synchrotron Radiation Powder X-ray Diffraction
High resolution in situ temperature varied synchrotron radiation powder X-ray diffraction (SR-PXD) experiments were performed at the Swiss Light Source (SLS), Switzerland and at the Diamond Light Source, Oxford, UK. At SLS, data were obtained at the Material Science powder diffraction beamline X04SA [34] using a monochromatic beam in Debye-Scherrer geometry with a Mythen microstrip detector with a wavelength of λ = 0.709396 Å. At Diamond, data were obtained at the I11 beamline [35] using a wide angle position sensitive detector and a wavelength of λ = 0.82646 Å. In both cases, the samples were loaded into 0.5 mm borosilicate glass capillaries in an Ar-filled glove box (<1 ppm of O 2 and H 2 O), sealed with grease and rotated during data acquisition. The samples were heated at 5 • C·min −1 from RT to 500 • C using a heat blower. The temperature was calibrated using a NaCl standard prior to diffraction runs [36]. The powder diffraction data were normalized and reduced, then modeled and refined according to the Rietveld method as implemented in the TOPAS software [37].
The structural models of Mg(NH 3 ) 6 Cl 2 , Mg(NH 3 ) 2 Cl 2 and Ni(NH 3 )Cl 2 were used as starting points for Rietveld refinements of the hexammine, diammine, and monoammine phases of the mixed metal halide ammines, respectively. The diffraction peaks were modeled by a Thompson-Cox-Hastings pseudo-Voigt function. The scale factor, zero-shift, unit cell parameters, atomic positions and background were refined. The N-H and H-H distances were restrained using soft restraints function during the Rietveld refinements.

Sorption Kinetics and Cycling
The two pristine materials MgCl 2 and MnCl 2 and the mixed metal halides Mg 0.9 Mn 0.1 Cl 2 and Mg 0.5 Mn 0.5 Cl 2 (m~0.1 g) were studied with regards to their NH 3 absorption and desorption kinetics. The absorption process was conducted under 2.5 bar of NH 3 at RT, while the desorption reaction was achieved by heating the samples up to 350 • C with a heating rate of 2 • C·min −1 under 1 bar of NH 3 . A calibrated volume consisting of a reference volume (V = 482.9 mL) and a sample chamber (V = 23.25 mL) was used during the experiments and the moles of absorbed and desorbed NH 3 were calculated according ideal gas law using the formula below: where ∆n is the number of NH 3 moles absorbed or desorbed, ∆P is the pressure change in the system occurring due to absorption or desorption of NH 3 , V is the volume, R is the gas constant and T is the temperature. In all cases, the number of absorbed or desorbed moles was normalized by the molar weight of the corresponding compound. Each NH 3 desorption was followed by evacuation of the released NH 3 from the cylinder to avoid reabsorption of the NH 3 gas during cooling to RT. The NH 3 desorption/absorption was cycled four times for each sample.  Table S1 in the Supporting Information. The diffraction patterns in Figure 1 confirm the formation of a MgCl 2 -MnCl 2 solid solutions, as only a single set of Bragg diffraction peaks belonging to Mg 1−x Mn x Cl 2 are observed, which is positioned in between that of MgCl 2 and MnCl 2 . MgCl 2 and MnCl 2 are isostructural and crystallize in the trigonal CdCl 2 -type structure with space group symmetry R-3m [38,39]. released NH3 from the cylinder to avoid reabsorption of the NH3 gas during cooling to RT. The NH3 desorption/absorption was cycled four times for each sample.

Structural Characterization of the As-Synthesized and Ammoniated Samples at RT
SR-PXD data were collected for the pristine samples, MgCl2 and MnCl2, and for the assynthesized Mg1-xMnxCl2, (x = 0.025, 0.05, 0.1, 0.3 and 0.5) samples. The unit cell parameters are presented in Table S1 in the Supporting Information. The diffraction patterns in Figure 1 confirm the formation of a MgCl2-MnCl2 solid solutions, as only a single set of Bragg diffraction peaks belonging to Mg1-xMnxCl2 are observed, which is positioned in between that of MgCl2 and MnCl2. MgCl2 and MnCl2 are isostructural and crystallize in the trigonal CdCl2-type structure with space group symmetry R-3m [38,39]. The MgCl2-MnCl2 solid solution follows Vegard's law approximately, as the volume is a function of the relative content of cations and in between that of the two pristine compounds, see Figure 2a. The larger ionic radius of Mn 2+ (0.83 Å) as compared to Mg 2+ (0.72 Å) results in an increase of the unit cell volume [40]. The deviation from Vegard's law might be due to a localized strain field caused by difference in Mg and Mn sizes, as well as to the different outer electronic structures of the mixing components (Mg and Mn in our case) [41]. The solid solution is maintained after ammoniation and the deviation from Vegard's law remains. It should be noted that a negative deviation from Vegard's law should be also expected if contamination from iron contained in the milling media results in the form Mg1−x-yMnxFeyCl2. Indeed, the ionic radius for Fe 2+ (0.63 Å) is smaller than the radii of both Mg 2+ and Mn 2+ [40]. However, even if slight contaminations from the milling media and metallic Fe cannot be ruled out, they are expected to be very limited due to the relatively short milling time (1 hour) used in this work. Additionally, metallic Fe must be oxidized to Fe 2+ in order to substitute Mg 2+ /Mn 2+ and form Mg1−x-yMnxFeyCl2. Therefore, it is most likely that the deviation from Vegard's law observed here might be due to the different outer electronic structures of Mg and Mn. Rietveld refinement and The MgCl 2 -MnCl 2 solid solution follows Vegard's law approximately, as the volume is a function of the relative content of cations and in between that of the two pristine compounds, see Figure 2a. The larger ionic radius of Mn 2+ (0.83 Å) as compared to Mg 2+ (0.72 Å) results in an increase of the unit cell volume [40]. The deviation from Vegard's law might be due to a localized strain field caused by difference in Mg and Mn sizes, as well as to the different outer electronic structures of the mixing components (Mg and Mn in our case) [41]. The solid solution is maintained after ammoniation and the deviation from Vegard's law remains. It should be noted that a negative deviation from Vegard's law should be also expected if contamination from iron contained in the milling media results in the form Mg 1−x−y Mn x Fe y Cl 2 . Indeed, the ionic radius for Fe 2+ (0.63 Å) is smaller than the radii of both Mg 2+ and Mn 2+ [40]. However, even if slight contaminations from the milling media and metallic Fe cannot be ruled out, they are expected to be very limited due to the relatively short milling time (1 h) used in this work. Additionally, metallic Fe must be oxidized to Fe 2+ in order to substitute Mg 2+ /Mn 2+ and form Mg 1−x−y Mn x Fe y Cl 2 . Therefore, it is most likely that the deviation from Vegard's law observed here might be due to the different outer electronic structures of Mg and Mn. Rietveld refinement and structural characterization of the ammoniated samples confirm the cubic Mg(NH 3 ) 6 Cl 2 structure for all mixed cation hexammines (Supporting Information, Figure S1-S5). Figure 2b illustrate Vegard's law (blue dotted line) for Mg 1−x Mn x (NH 3 ) 6 Cl 2 . Surprisingly, the unit cell volume for samples with x < 0.05 are lower than that of Mg(NH 3 ) 6 Cl 2 .
Energies 2020, 13, x FOR PEER REVIEW 5 of 15 law (blue dotted line) for Mg1-xMnx(NH3)6Cl2. Surprisingly, the unit cell volume for samples with x < 0.05 are lower than that of Mg(NH3)6Cl2. The atomic positions of the monometallic and mixed hexammines obtained from Rietveld refinement are presented in Table 1  0.21540 (14), 0, 0 * The data are obtained at RT.    Figure 3 shows the TGA-DSC measurements performed on Mg(NH 3 ) 6 Cl 2 , Mn(NH 3 ) 6 Cl 2 and Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 . The DSC measurements for the other mixed ammines are shown in the Supporting Information ( Figure S6). All the hexammine compounds are relatively stable at RT, except for Mn(NH 3 ) 6 Cl 2 , which slowly releases NH 3 in the glove box at RT. Thus, Mn(NH 3 ) 6 Cl 2 was kept at T = −34 • C in a glovebox freezer prior to the TGA-DSC measurements.  Each NH3 desorption step is followed by mass loss. In the first desorption step, 4 NH3 moles are released and Mg0.5Mn0.5(NH3)6Cl2 experience a 30.2% mass loss, while the next two desorption events reduce the mass of the sample by 8.3% and 8.1%, respectively. The mass loss ratio 4:1:1 of the monometallic and mixed hexammines (Supporting Information, Table S2) corresponds to the moles of NH3 desorbed in each desorption event, i.e., four moles of NH3 released in the first desorption step, and 1 mole of NH3 released in the second and third step, respectively, and agrees well with the theoretical weight loss expected from the NH3 desorption. The gravimetric NH3 capacities for the monometallic and the mixed cation hexammines are presented in Table S3. SR-PXD data measured of Mg1-xMnx(NH3)6Cl2, (x = 0, 0.025, 0.05, 0.1, 0.3 and 0.5) after the TGA-DSC measurements confirm the reformation of Mg1-xMnxCl2 after full NH3 release, thus confirming the stability of the solid solution (Supporting Information, Figure S7).

Thermal Analysis
Kissinger analysis was performed on the DSC heat flow signals for the three desorption events measured for Mg(NH3)6Cl2, Mn(NH3)6Cl2 and Mg0.5Mn0.5(NH3)6Cl2 to determine the activation energy and investigate their NH3 desorption kinetics. Kissinger plots for the three endothermic events with the release of 4, 1 and 1 moles of NH3 are shown in Figure 4a-c. The corresponding activation energies were calculated for each desorption event and are presented in Figure 4d-f. The activation energy of the first four moles of NH3 desorption from Mg0.5Mn0.5(NH3)6Cl2 is 54.8 kJ⋅mol -1 , which is approximately in between that of Mg(NH3)6Cl2 (Ea = 60.8 kJ⋅mol -1 ) and Mn(NH3)6Cl2 (Ea = 43.5 kJ⋅mol -1 ). For the second NH3 desorption step, the activation energy for Mg0.5Mn0.5(NH3)2Cl2 is 73.2 kJ⋅mol -1 , in between that of Mg(NH3)2Cl2 (Ea = 74.8 kJ⋅mol -1 ) and Mn(NH3)2Cl2 (Ea = 67.7 kJ⋅mol -1 ). The final desorption event of Mg0.5Mn0.5(NH3)Cl2 has an activation energy of 91.0 kJ⋅mol -1 , as compared to Mg(NH3)Cl2 (Ea = 91.8 kJ⋅mol -1 ) and Mn(NH3)Cl2 (Ea = 90.9 kJ⋅mol -1 ). Each NH 3 desorption step is followed by mass loss. In the first desorption step, 4 NH 3 moles are released and Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 experience a 30.2% mass loss, while the next two desorption events reduce the mass of the sample by 8.3% and 8.1%, respectively. The mass loss ratio 4:1:1 of the monometallic and mixed hexammines (Supporting Information, Table S2) corresponds to the moles of NH 3 desorbed in each desorption event, i.e., four moles of NH 3 released in the first desorption step, and 1 mole of NH 3 released in the second and third step, respectively, and agrees well with the theoretical weight loss expected from the NH 3 desorption. The gravimetric NH 3 capacities for the monometallic and the mixed cation hexammines are presented in Table S3. SR-PXD data measured of Mg 1−x Mn x (NH 3 ) 6 Cl 2 , (x = 0, 0.025, 0.05, 0.1, 0.3 and 0.5) after the TGA-DSC measurements confirm the reformation of Mg 1−x Mn x Cl 2 after full NH 3 release, thus confirming the stability of the solid solution (Supporting Information, Figure S7).
Kissinger analysis was performed on the DSC heat flow signals for the three desorption events measured for Mg(NH 3 ) 6  The activation energies for Mg0.5Mn0.5(NH3)6Cl2 in all three desorption events are significantly lower as compared to monometallic Mg(NH3)6Cl2. Therefore, by obtaining the mixed metal halide ammines, it is possible to tailor the desorption temperature and kinetics of the mixed metal halide ammines compared to the monometallic halide ammines.

In Situ SR-PXD
The in situ SR-PXD data for Mg0.5Mn0.5(NH3)6Cl2 in the temperature range RT to 402 °C, with a heating rate of 5 °C⋅min -1 , are shown in Figure 5a, while Figure 5b shows diffraction patterns at selected temperatures for each of the ammoniated compounds observed during heating. Rietveld refinements of the mixed metal hexammines are presented in the supporting material ( Figures S1-S5). The in situ SR-PXD data from RT for monometallic Mg(NH3)6Cl2 is shown in the supporting material ( Figure S6). SR-PXD data at RT contain Bragg peaks from Mg0.5Mn0.5(NH3)6Cl2 (97.6(6) wt%) The activation energies for Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 in all three desorption events are significantly lower as compared to monometallic Mg(NH 3 ) 6 Cl 2 . Therefore, by obtaining the mixed metal halide ammines, it is possible to tailor the desorption temperature and kinetics of the mixed metal halide ammines compared to the monometallic halide ammines.

In Situ SR-PXD
The in situ SR-PXD data for Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 in the temperature range RT to 402 • C, with a heating rate of 5 • C·min −1 , are shown in Figure 5a, while Figure 5b shows diffraction patterns at selected temperatures for each of the ammoniated compounds observed during heating. Rietveld refinements of the mixed metal hexammines are presented in the supporting material ( Figures S1-S5). The in situ SR-PXD data from RT for monometallic Mg(NH 3 ) 6 Cl 2 is shown in the supporting material ( Figure S6). SR-PXD data at RT contain Bragg peaks from Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 (97.6(6) wt%) and Mg 0.5 Mn 0.5 (NH 3 ) 2 Cl 2 (2.4(5) wt%). Upon heating, the Bragg peaks corresponding to Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 disappear between 115 and 125 • C, while peaks corresponding to Mg 0.5 Mn 0.5 (NH 3 ) 2 Cl 2 increase significantly in intensity. Upon further heating, the Bragg peaks corresponding to Mg 0.5 Mn 0.5 (NH 3 ) 2 Cl 2 decrease in intensity from~230 • C and peaks from Mg 0.5 Mn 0.5 (NH 3 )Cl 2 appear at~246 • C. The peaks from Mg 0.5 Mn 0.5 (NH 3 )Cl 2 disappear at 325 • C and some Bragg peaks from an unknown compound appear at 328 • C (Figure 5b, yellow diffraction pattern) before the full desorption of NH 3 and formation of the Mg 0.5 Mn 0.5 Cl 2 solid solution. The appearance of unknown diffraction peaks might be due to a non-stoichiometric transition from the Mg 0.5 Mn 0.5 (NH 3 )Cl 2 monoammine to the Mg 0.5 Mn 0.5 Cl 2 chloride. Such behavior was previously reported for Mn(NH 3 )Cl 2 , which was observed to release the last NH 3 via two or more desorption events [13,42]. This is also observed in our data from the Sieverts measurements and will be discussed later in Section 3.4.
In situ SR-PXD data for Mn(NH 3 ) 6 Cl 2 in the temperature range RT to 406 • C, with a heating rate of 5 • C·min −1 confirms the presence of such intermediate phase at 307 • C ( Figure S9). However, due to the fast heating rate and, therefore, dominant peaks from Mn(NH 3 )Cl 2 and MnCl 2 phases in the diffraction pattern, it was challenging to index it and extract the unit cell parameters for Mn(NH 3 ) 1-δ Cl 2 . The same applies for the diffraction pattern of possible Mg 0.5 Mn 0.5 (NH 3 ) 1-δ Cl 2 phase at 328 • C from the in situ data for Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 ( Figure 5), where the dominant diffraction peaks from Mg 0.5 Mn 0.5 (NH 3 )Cl 2 make the indexing challenging. Observation of these intermediate diffraction patterns suggests that the transition of Mn(NH 3 )Cl 2 to MnCl 2 and Mg 0.5 Mn 0.5 (NH 3 )Cl 2 to Mg 0.5 Mn 0.5 Cl 2 , which was described as non-stoichiometric process [42], in our study undergoes by stoichiometric NH 3 releases of δ moles. Mg0.5Mn0.5(NH3)2Cl2 increase significantly in intensity. Upon further heating, the Bragg peaks corresponding to Mg0.5Mn0.5(NH3)2Cl2 decrease in intensity from ~230 °C and peaks from Mg0.5Mn0.5(NH3)Cl2 appear at ~246 °C. The peaks from Mg0.5Mn0.5(NH3)Cl2 disappear at 325 °C and some Bragg peaks from an unknown compound appear at 328 °C (Figure 5b, yellow diffraction pattern) before the full desorption of NH3 and formation of the Mg0.5Mn0.5Cl2 solid solution. The appearance of unknown diffraction peaks might be due to a non-stoichiometric transition from the Mg0.5Mn0.5(NH3)Cl2 monoammine to the Mg0.5Mn0.5Cl2 chloride. Such behavior was previously reported for Mn(NH3)Cl2, which was observed to release the last NH3 via two or more desorption events [13,42]. This is also observed in our data from the Sieverts measurements and will be discussed later in Section 3.4.
The monoammine Mg0.5Mn0.5(NH3)Cl2 is isostructural to Ni(NH3)Cl2, which crystallizes in a monoclinic unit cell with space group I2/m [44] where each Cl atom is shared by three Mg0.5Mn0.5 atoms in edge-sharing double octahedral chains. Both broad and narrow diffraction peaks are observed for the monoammine phase (see Figure 6c), indicating the presence of structural disorder or stacking faults in the structure, resulting in a marked difference between the experimental and calculated patterns.
The Rietveld refinement of the fully desorbed Mg0.5Mn0.5Cl2 at 402 °C is shown in Figure 6d. Mg0.5Mn0.5Cl2 structure, space group R-3m, is formed by the octahedra of Cl atoms with central Mg0.5Mn0.5 atoms sharing half of their edges, and thus resulting in layers of Mg0.5Mn0.5Cl2. Rietveld refinement of Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 at RT is performed using the structural model of Mg(NH 3 ) 6 Cl 2 (Figure 6a). Two phases are present in the sample, which are identified as Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 and Mg 0.5 Mn 0.5 (NH 3 ) 2 Cl 2 with the refined phase fractions to 97.6(6) wt% and 2.4(5) wt%, respectively, which might be resulted from partial NH 3 release at RT. Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 crystallizes in a cubic unit cell, a = 10.22037(8) Å at RT with space group Fm-3m and is isostructural to Mg(NH 3 ) 6 Cl 2 . Octahedral Mg 0.5 Mn 0.5 (NH 3 ) 6 complexes are contained in a cubic lattice of Cl atoms, with each Mg 0.5 Mn 0.5 atom octahedrally coordinated by six N atoms.
The monoammine Mg 0.5 Mn 0.5 (NH 3 )Cl 2 is isostructural to Ni(NH 3 )Cl 2 , which crystallizes in a monoclinic unit cell with space group I2/m [44] where each Cl atom is shared by three Mg 0.5 Mn 0.5 atoms in edge-sharing double octahedral chains. Both broad and narrow diffraction peaks are observed for the monoammine phase (see Figure 6c), indicating the presence of structural disorder or stacking faults in the structure, resulting in a marked difference between the experimental and calculated patterns.
The Rietveld refinement of the fully desorbed Mg 0.5 Mn 0.5 Cl 2 at 402 • C is shown in Figure 6d. Mg 0.5 Mn 0.5 Cl 2 structure, space group R-3m, is formed by the octahedra of Cl atoms with central Mg 0.5 Mn 0.5 atoms sharing half of their edges, and thus resulting in layers of Mg 0.5 Mn 0.5 Cl 2 .

NH 3 Cycling and Kinetics
Studies of the NH 3 sorption kinetics and cyclability of the pristine and mixed metal halides with the lower and higher Mn contents (Mg 0.9 Mn 0.1 Cl 2 and Mg 0.5 Mn 0.5 Cl 2 ) were performed using a Sieverts apparatus. The results from the desorption cycles performed on Mg 0.9 Mn 0.1 (NH 3 ) 6 Cl 2 are presented in the supporting information ( Figure S10). The fifth NH 3 absorption process (after four cycles) of pristine MgCl 2 , MnCl 2 and Mg 0.5 Mn 0.5 Cl 2 are presented in Figure 7. For absorption, the applied NH 3 gas pressure was~2.5 bar and the processes were conducted at RT. The moles of absorbed NH 3 were calculated (see Equation (1)), where ∆n is calculated from the pressure drop, ∆P, occurring during NH 3 absorption. The observed pressure drop was ∆P = 0.33 bar and the final pressures at the end of absorption were 2.21 and 2.22 bar for MgCl 2 and Mg 0.5 Mn 0.5 Cl 2 , respectively. MnCl 2 only absorbed 5.5 moles of NH 3 , which corresponds to a pressure drop of only ∆P = 0.29 bar reaching a final pressure p = 2.23 bar. This indicated that not all the MnCl 2 powder had reacted with NH 3 , despite the still relatively high value of the final pressure of absorption. Indeed, due to the large volume expansion of the metal chloride during ammonia absorption, some clogging may occur and prevent ammonia from reaching all the salt crystals [16]. On the other, it cannot be excluded that that the absorption reaction stops because the equilibrium pressure for Mn(NH 3 ) 6 Cl 2 at RT is higher than the final pressure reached during absorption (p = 2.23 bar). A more detailed thermodynamic study using pressure-composition-isotherms is needed to clarify this in detail. MgCl 2 absorbed 6 moles of NH 3 in less than 1000 s, while Mg 0.5 Mn 0.5 Cl 2 absorbed 6 moles of NH 3 in 6000 s. Similarly, the NH 3 absorption rate for the pristine halides are very different: Four moles of NH 3 is absorbed in MgCl 2 in about 200 s, while it took about 800 s to absorb the similar amount of NH 3 in MnCl 2 . The rate of absorption for Mg 0.5 Mn 0.5 Cl 2 is similar to that of MnCl 2 , indicating that Mn plays a predominant role for governing the kinetics of the hexammine formation.
The NH 3 desorption processes during cycling were carried out upon heating with a constant heating rate of 2 • C·min −1 from RT to 350 • C under an initial NH 3 pressure of 1 bar, see Figure 8 . The moles of desorbed NH 3 were calculated using Equation (1) and the pressure increase, ∆P = 0.32 bar, due to NH 3 release. The three desorption steps of NH 3 were observed as a pressure increase and ∆n was calculated. For Mg 0.5 Mn 0.5 (NH 3 ) 2 Cl 2 , the first 4 moles of NH 3 starts desorbing at around 100 • C and are fully released at 166 • C. The resulting Mg 0.5 Mn 0.5 (NH 3 ) 2 Cl 2 desorbs one mole of NH 3 in the temperature range from 240 • C to 260 • C, forming Mg 0.5 Mn 0.5 (NH 3 )Cl 2 . The final NH 3 desorption step occurs above 300 • C. However, the transformation from monoammine to fully desorbed mixed metal chloride, Mg 0.5 Mn 0.5 Cl 2 , does not proceed via a single step as for the previous desorption. Instead it undergoes through two discrete steps, consistent with the observations of a different crystalline phase in the in situ SR-PXD experiments. The NH3 desorption processes during cycling were carried out upon heating with a constant heating rate of 2 °C⋅min -1 from RT to 350 °C under an initial NH3 pressure of 1 bar, see Figure 8. The moles of desorbed NH3 were calculated using Equation 1 and the pressure increase, ΔP = 0.32 bar, due to NH3 release. The three desorption steps of NH3 were observed as a pressure increase and Δn was calculated. For Mg0.5Mn0.5(NH3)2Cl2, the first 4 moles of NH3 starts desorbing at around 100 °C and are fully released at 166 °C. The resulting Mg0.5Mn0.5(NH3)2Cl2 desorbs one mole of NH3 in the temperature range from 240 °C to 260 °C, forming Mg0.5Mn0.5(NH3)Cl2. The final NH3 desorption step occurs above 300 °C. However, the transformation from monoammine to fully desorbed mixed metal chloride, Mg0.5Mn0.5Cl2, does not proceed via a single step as for the previous desorption. Instead it undergoes through two discrete steps, consistent with the observations of a different crystalline phase in the in situ SR-PXD experiments. For some hexammines, M(NH3)6Cl2 (M = Mg, Ni), the desorption consists of three events [13]. In contrast, Mn(NH3)6Cl2 was reported to undergo non-stoichiometric NH3 release in the last desorption step [13,42]. In our study, we observe two separate steps, and from the number of desorbed NH3 moles calculated from ΔP in Sieverts studies, δ seems to be equal to 0.5. Furthermore, the NH3 desorption studies of Mg0.9Mn0.1(NH3)6Cl2 ( Figure S10) suggest that this phenomenon does not occur in Mg1-xMnx(NH3)6Cl2 with low Mn content, as the last decomposition step occurs as a single event.  The NH3 desorption processes during cycling were carried out upon heating with a constant heating rate of 2 °C⋅min -1 from RT to 350 °C under an initial NH3 pressure of 1 bar, see Figure 8. The moles of desorbed NH3 were calculated using Equation 1 and the pressure increase, ΔP = 0.32 bar, due to NH3 release. The three desorption steps of NH3 were observed as a pressure increase and Δn was calculated. For Mg0.5Mn0.5(NH3)2Cl2, the first 4 moles of NH3 starts desorbing at around 100 °C and are fully released at 166 °C. The resulting Mg0.5Mn0.5(NH3)2Cl2 desorbs one mole of NH3 in the temperature range from 240 °C to 260 °C, forming Mg0.5Mn0.5(NH3)Cl2. The final NH3 desorption step occurs above 300 °C. However, the transformation from monoammine to fully desorbed mixed metal chloride, Mg0.5Mn0.5Cl2, does not proceed via a single step as for the previous desorption. Instead it undergoes through two discrete steps, consistent with the observations of a different crystalline phase in the in situ SR-PXD experiments. For some hexammines, M(NH3)6Cl2 (M = Mg, Ni), the desorption consists of three events [13]. In contrast, Mn(NH3)6Cl2 was reported to undergo non-stoichiometric NH3 release in the last desorption step [13,42]. In our study, we observe two separate steps, and from the number of desorbed NH3 moles calculated from ΔP in Sieverts studies, δ seems to be equal to 0.5. Furthermore, the NH3 desorption studies of Mg0.9Mn0.1(NH3)6Cl2 ( Figure S10) suggest that this phenomenon does not occur in Mg1-xMnx(NH3)6Cl2 with low Mn content, as the last decomposition step occurs as a single event. For some hexammines, M(NH 3 ) 6 Cl 2 (M = Mg, Ni), the desorption consists of three events [13]. In contrast, Mn(NH 3 ) 6 Cl 2 was reported to undergo non-stoichiometric NH 3 release in the last desorption step [13,42]. In our study, we observe two separate steps, and from the number of desorbed NH 3 moles calculated from ∆P in Sieverts studies, δ seems to be equal to 0.5. Furthermore, the NH 3 desorption studies of Mg 0.9 Mn 0.1 (NH 3 ) 6 Cl 2 ( Figure S10) suggest that this phenomenon does not occur in Mg 1−x Mn x (NH 3 ) 6 Cl 2 with low Mn content, as the last decomposition step occurs as a single event. This indicates that sufficiently high amount of Mn in Mg 1−x Mn x (NH 3 ) 6 Cl 2 results in a change in the physical behavior to be similar to that of Mn(NH 3 ) 6 Cl 2 .
These results suggest that by changing the relative Mg:Mn ratio in Mg 1−x Mn x (NH 3 ) 6 Cl 2 the NH 3 sorption propertied can be tuned and optimized. For instance, substituting Mg in Mn(NH 3 ) 6 Cl 2 increases its stability, avoiding NH 3 desorption at RT. Furthermore, due to the low weight of Mg, the gravimetric capacity increases, with increasing Mg content. Finally, increasing the relative content of magnesium can be beneficial if cost reduction is desirable.
A thorough investigation of the NH 3 desorption reaction enthalpies is planned for further thermodynamic studies of the Mg 1−x Mn x (NH 3 ) 6 Cl 2 hexammines by applying pressure composition isotherm (PCI) studies.

Conclusions
A series of novel mixed metal halide ammines, Mg 1−x Mn x (NH 3 ) 6 Cl 2 , with a usable ammonia capacity in the temperature range 20-350 • C were synthesized and characterized. The crystal structures of the different ammine phases are identified and investigated by in situ SR-PXD. All Mg 1−x Mn x (NH 3 ) 6 Cl 2 solid solutions crystallize in a cubic unit cell with space group symmetry Fm-3m and unit cell parameters intermediate that of the two monometallic materials, Mg(NH 3 ) 6 Cl 2 and Mn(NH 3 ) 6 Cl 2 . DSC analysis reveal a decrease in the onset temperature for NH 3 desorption for the solid solutions as compared to the monometallic Mg(NH 3 ) 6 Cl 2 . Activation energies for each desorption step are calculated and show the possibility of tailoring the activation energies for the NH 3 release in mixed metal chloride hexammines. The lower activation energies for NH 3 desorption in Mn(NH 3 ) 6 Cl 2 resulted in a lowering of the activation energies for the solid solution Mg 0.5 Mn 0.5 (NH 3 ) 6 Cl 2 . Finally, NH 3 reversibility measurements reveal that the solid solution has a high stability, thus making them promising candidates for solid-state NH 3 storage systems.