Stabilization of Mn⁴⁺ in Synthetic Slags and Identification of Important Slag Forming Phases

Abstract

The expected shortage of Li due to the strong increase in electromobility is an important issue for the recovery of Li from spent Li-ion batteries. One approach is pyrometallurgical processing, during which ignoble elements such as Li, Al and Mn enter the slag system. The engineered artificial mineral (EnAM) strategy aims to efficiently recover critical elements. This study focuses on stabilizing Li-manganates in a synthetic slag and investigates the relationship between Mn⁴⁺ and Mg and Al in relation to phase formation. Therefore, three synthetic slags (Li, Mg, Al, Si, Ca, Mn, O) were synthesized. In addition to LiMn³⁺O₂, Li₂Mn⁴⁺O₃ was also stabilized. Both phases crystallized in a Ca-silicate-rich matrix. In the structures of Li₂MnO₃ and LiMnO₂, Li and Mn can substitute each other in certain proportions. As long as a mix of Mn²⁺ and Mn³⁺ is present in the slag, spinels form through the addition of Mg and/or Al.

1. Introduction

1.1. Recovery of Critical Elements from the Slag Phase

The usage of critically relevant elements, such as Li, will be increasingly in demand due to their use in various technological areas [1]. The EU has established a list of critical elements, which includes Li as well as Co [2]. In Li-ion batteries (LIBs), these two elements are used as cathode materials (LiCoO₂). The geological supply of such elements is limited, thus the recovery of, e.g., Li from the industrial waste stream. such as from old LIBs, is indispensable [3,4,5]. Pyrometallurgical processes [6] are a promising method for the recovery. With this route, various elements (e.g., Co, Ni or Cu) can be recovered directly, while base elements (e.g., Li, Mg, Al or Mn) are either lost as dust or enter the slag phase and form complex compounds. For efficient recovery of critical elements such as Li, it is essential to concentrate the element of interest in a single phase within the slag [1,7]. In addition, this phase should have a high Li-content as well as good processing properties (e.g., habitus, crystal size, magnetic properties). The stabilization of Li in a single phase for an efficient recovery of critically relevant elements is the idea behind engineered artificial minerals (EnAMs) [8] (Figure 1).

1.2. Potential EnAMs for Efficient Recovery

A study by Elwert et al. [9] shows that the Li-aluminate LiAlO₂ would be suitable as a potential EnAM, as the idiomorphic to hypidiomorphic crystals can be separated from the remaining slag via flotation [10]. In the presence of especially Mg, Al and Mn, spinel complexes are formed during solidification and solid solutions occurred between spinels (e.g., MgMn₂O₄, MgAl₂O₄, MnAl₂O₄ and Mn²⁺Mn³⁺₂O₄), which hampered the formation of LiAlO₂ [11,12,13]. In addition, this Li aluminate can contain up to 3 wt.% Si, which complicates hydrometallurgical processing [10]. As the use of Mn in new Li batteries increases, Mn will become part of a complex slag system along with Li, Mg and Al. In addition, small amounts of Mg and Al in LiMnO₂ and Li₂MnO₃ are expected to promote intralayer diffusion and Mg should also promote interlayer diffusion [14]. For this reason, current research is focusing on the stabilization of Li in Li-manganates as a potential EnAM [8]. The difficulties here are the redox-sensitive behavior of Mn and the Jahn–Teller effect on the Mn³⁺ at temperatures above 1445 K [12,13]. Schnickmann et al. [8] showed that it is possible to stabilize a Li-manganate in a complex synthetic oxide slag system (Li, Mg, Al, Si, Ca, Mn). Under a normal atmosphere, a mixture between Mn²⁺ and Mn³⁺ was present and pure LiMnO₂ was formed. Nevertheless, the formation of spinels and spinel solid solutions could not be prevented, as well as the incorporation of Mn²⁺ into the Ca-silicate matrix. In a follow-up experiment, pure oxygen (100% O₂) was used to stabilize a higher Mn oxidation stage in the slag. For this experiment, the same batch precursors were used as in Schnickmann et al. [8]. This experiment is designed to investigate whether it is possible to stabilize higher Mn oxidation states in synthetic slags and whether Mn⁴⁺ forms a compound with the other elements, especially with Mg and/or Al.

1.3. Important Slag Forming Phases

In a complex slag system consisting of Li₂O, MgO, Al₂O₃, SiO₂, CaO and MnO, mainly binary and ternary oxide compounds are formed, followed by a residual melt [8]. The compounds described below represent possible phases in a crystalline slag with a special focus on Li-rich compounds and the Mn oxides. Compounds between Mn⁴⁺ and Al/Mg have not yet been described in the literature. The ternary compounds are of minor importance for the results presented and are therefore not discussed in detail in this chapter.

The Li₂O-MnO_x system (1 ≤ x ≤ 2) must be considered for the investigation of possible Mn-containing EnAMs. Due to the redox sensitivity of Mn, the formation of various Li-manganates with different concentrations and Mn speciations are possible. The phase equilibria in the Li₂O-MnO_x system in air is shown in Figure 2. According to the experimental data obtained by Paulsen and Dahn [15], cubic spinel transforms to tetragonal spinel at high temperatures. A miscibility gap between hausmannite (Mn₃O₄) and t-spinel was proposed. LiMnO₂ is stable at temperatures higher than 1223 K, while the Li₂MnO₃ phase is stable up to approximately 1240 K. In addition to Paulsen and Dahn [15], Longo et al. [16] also described such compounds in detail, as well as the influence of temperature, pressure and pH on the formation and stability of such Li-manganates. Below 400 °C, the cubic spinel phases with stoichiometry, LiMn_1.75O₄ and Li₄Mn₅O₁₂, are found to be stable. They can also be part of the spinel solid solution, as shown in the calculations of [17], but to achieve equilibrium at these low temperatures would be problematic. The phase Li₂Mn₃O₇ is calculated to be stable, but its stability has not yet been experimentally proven [15,16,18]. Between 400 and 800 °C, the Li-spinel (Li_(1+x)Mn_(2−x)O₄) has a large stability field between LiMn₂O₄ (Mn^3.5+) and Li₄Mn₅O₁₂ (Mn⁴⁺) and can coexist with Li₂MnO₃ (Mn⁴⁺). Above 800 °C, Li₂MnO₃ and LiMn₂O₄ decompose into Mn₃O₄ and O₂ alongside layered LiMnO₂. From 1000 to 1060 °C, orthorhombic LiMnO₂ can coexist with lithiated hausmannite (Li_xMn_3−xO₄) [15]. According to Mishra and Ceder [19], the Jahn–Teller distortion affects the tetragonal spinel LiMn₂O₄ as well as orthorhombic (e.g., LiMnO₂) and monoclinic layered structures [19]. In addition, the investigations by Schnickmann et al. [8] showed that in the LiMnO₂ system the Li and Mn contents can be exchanged and that this compound can incorporate up to 0.35 wt.% Al.

Due to their good flotation properties, Li-aluminates have been investigated as potential EnAMs for a long time. Recently, thermodynamic assessments of the Li₂O-Al₂O₃ system using the CALPHAD approach have been published by Konar et al. [20] and De Abreu et al. [21]. Moreover, De Abreu et al. experimentally studied the phase equilibria in the Al₂O₃-rich part of the diagram and made calorimetric measurements of heat capacity for intermediate phases. The calculated phase diagram is shown in Figure 3a. Three stable intermediate phases were proposed: LiAl₅O₈, LiAlO₂ and Li₅AlO₄. LiAlO₂ has already been extensively analyzed and discussed for its potential as an EnAM [20,21]. Two polymorphs are reported for LiAlO₂: the low-temperature α-phase (trigonal) and the high-temperature γ-phase (tetragonal). Two stable modifications of LiAl₅O₈ with spinel structure but with different space groups were found in the alumina-rich side of the phase diagram. High-temperature modification of the spinel is inversed and stable with some homogeneity range. No evidence of LiAl₁₁O₁₇ was found in the microstructures and thus LiAl₁₁O₁₇ is not treated as stable.

Another important area of Li-rich phases is the Li₂O-SiO₂ system, with the following stable phases: Li₂Si₂O₅, Li₂SiO₃, Li₄SiO₄ and Li₈SiO₆ (ordered by decreasing Si concentration). According to the latest results, the Li₆Si₂O₇ phase is treated as metastable. Li₂SiO₃ is formed from a Li₂O-SiO₂-rich melt.

Spinels are the most important Li-free compounds, whereas hausmannite (Mn²⁺Mn³⁺₂O₄) is the most important compound in the MnO-Mn₂O₃ system. Below 1172 °C, the Jahn–Teller effect has an influence on the Mn³⁺ position and causes deformation of the crystal parameters. The transformation from tetragonal to cubic is reversible and crystal lattice changes from cubic to tetragonal on cooling. Hausmannite can be regarded as a low-temperature phase as well as a deformed spinel under normal conditions (1 atm; 25 °C) and low oxygen partial pressure. Hausmannite forms from Mn₂O₃ at 1445 K and air oxygen partial pressure. It can form solid solutions with other spinel compounds. One of the most important spinels in the MnO-Al₂O₃ system is galaxite (MnAl₂O₄) [12,13,20]. The calculated phase diagrams with thermodynamic parameters optimized by [13] for the Al₂O₃-MnO_x system in air and in the presence of metallic Mn are shown in Figure 3b,c, respectively. In the presence of metallic Mn, thus in varying oxygen partial pressure p(O₂) with temperature, the homogeneity range of the spinel phase is narrow and its stoichiometry is close to MnAl₂O₄. According to crystallographic data, the spinel phase is normal at low temperature, with tetrahedral sites occupied by Mn²⁺ cations, but with temperature increases the degree of inversion (fraction of Al³⁺ in tetrahedral sites) increases. According to calculations, which agree with experimental results, the homogeneity range of spinel extends with an increase in oxygen partial pressure and in air conditions (p(O₂) = 0.21 bar) the homogeneity range of the cubic spinel extends from Mn₃O₄ to MnAl₂O₄ (at high temperatures). It should be noted that the Mn₃O₄ spinel forming from Mn₂O₃ via oxygen release has a tetragonal structure. The tetrahedral sites are occupied by Mn²⁺ and the octahedral sites by Mn³⁺. Tetragonal spinel transforms to cubic types at 1443 K, probably due to the disproportionation reaction Mn³⁺ → Mn²⁺ + Mn⁴⁺ to reduce the Jahn–Teller distortion caused by Mn³⁺.

Phase diagrams of other important systems containing the spinel phase, like MgO-MnO_x and MgO-Al₂O₃ systems are also shown in Figure 3d,e, respectively. The thermodynamic parameters obtained by [22] and [23] were applied to calculate the phase relationships for these binary systems. It should be noted that spinels with tetragonal and cubic structures were found in the MgO-MnOx system. Tetrahedral spinel has a homogeneity range from Mn₃O₄ to MgMn₂O₄, while cubic spinel extends from Mn₃O₄ to Mg₂MnO₄.

2. Materials and Methods

2.1. Chemicals and Preparation of Precursors and Reference Materials

For the melting experiment, the same batch of precursors was used as in the study by Schnickmann et al. [9], with the following chemical compositions: LiNO₃, Mg(NO₃)₂ × 6 H₂O, Al(NO₃)₃ × 9 H₂O, Köstrosol^® 0830 AS, Ca(NO₃)₂ × 4 H₂O, Mn(NO₃)₂ × 4 H₂O and C₆H₈O₇.

The gel-combustion method was used to prepare the precursors. For this purpose, common procedures [24,25] were adapted to the experimental setup in the laboratory. The method has also been described by Schnickmann et al. [8]. The gel-combustion product was thermally treated with NH₄NO₃ in quartz crucibles up to 480 °C (10 °K/min; Nabertherm LE 1/11/R7, Nabertherm GmbH, Lilienthal, Germany) to remove the residual carbon. In

Table 1. Elemental composition of the precursors in wt.% and at.%. The precursors have the same composition as [8].

	Precursor 1		Precursor 2		Precursor 3
Element	wt.%	at.%	wt.%	at.%	wt.%	at.%
Li	1.9	6.3	1.9	6.3	1.8	5.8
Mg	0.0	0.0	1.2	1.1	1.2	1.1
Al	0.0	0.0	0.0	0.0	1.1	0.9
Si	19.3	15.9	19.1	15.6	18.9	15.4
Ca	29.9	17.2	29.4	16.9	29.0	16.7
Mn	9.8	4.1	9.2	3.9	8.6	3.7
O	39.1	56.4	39.2	56.2	39.4	56.4
Total	100.0	100.0	100.0	100.0	100.0	100.0

, the elemental composition of the precursors can be found.

2.2. Melting Program

For the melting experiment, the thermally treated precursors were melted in Pt/Rh crucibles (filling capacity: 15 mL) under pure oxygen atmosphere at 200 L per hour in a chamber furnace (Nabertherm HT16/17, Nabertherm GmbH, Lilienthal, Germany).

The melt program of Schirmer et al. [10] and Schnickmann et al. [8] was adapted for this process (Figure 4). For each experiment, 1.20 g of sample material was used. The powder was not pressed or compacted in the crucible. All three samples were melted together in one run, in different crucibles, to achieve consistent conditions.

Due to the reaction-resistant material, no reaction took place between the slag and the crucible. However, not all of the slag material could be removed from the crucible, resulting in the loss of small amounts. For example, 1.03 g of S1, 1.01 g of S2 and 1.05 g of S3 were present after the melting experiment. The areas of slag near the surface took on a reddish color, while the rest of the slag was dark gray to black. EPMA was used to analyze the near-surface samples and the remaining slag, which is representative of the entire slag.

2.3. Characterization Techniques

The mineralogical composition of the slags was quantitatively and qualitatively identified using powder X-ray diffraction (PXRD) and electron probe microanalysis (EPMA).

The crystalline phases in the slag sample were analyzed using PXRD (Panalytical X-Pert Pro Diffractometer, Malvern Panalytical GmbH, Kassel, Germany). For the measurement (2θ-angle range from 5° to 100°; step size 0.0066°; 150.45 s per step), a Co-X-ray tube (Malvern Panalytical GmbH, Kassel, Germany, λ = 1.7902 Å, 40 kV, 40 mA) was used. The individual crystalline phases were classified via the American Mineralogist Crystal Structure Database [26] and the pdf-2 ICCD XRD database [27].

The elemental composition of single grains/crystals in the prepared thin sections were determined with EPMA (Cameca SXFIVE FE Field Emission, CAMECA SAS, Gennevilliers Cedex, France) using the Kα lines (Mg, Al, Si, Ca, Mn). For the measurement (15 kV beam diameter 100–600 nm; Schottky type [28]), the device was calibrated beforehand with certified reference materials (CRM: P&H Developments Ltd.; Glossop, Derbyshire, UK and Astimex Standards Ltd.; Toronto, ON, Canada). The measured intensities of the emitted X-rays were evaluated using the X-PHI model [29]. With the chosen method, the Li content cannot be analyzed quantitatively with the required precision. Therefore, the element concentration of Li was determined using virtual compounds according to T. Schirmer [30].

In order to validate the measurement method and obtain consistent results, repeated measurements were carried out on different days using the international standard rhodonite (MnSiO₃; Astimex). The low standard deviation (0.03) of Mn indicates that this element content can be analyzed very precisely and is suitable for calculating the Li content via virtual compounds (

Table 2. The CRM rhodonite was used to verify the measurement accuracy of Mn. The Mn content can be determined with an inaccuracy of ±0.03%. Indication in wt.%. * The iron content was not measured.

Wt.%	Average Rhodonite	%StDev. Rhodonite	Ref. Rhodonite
MgO	1.92	0.01	2.0
Al2O3	0.011	0.002	n.a.
SiO2	46.76	0.06	46.8
CaO	4.74	0.04	4.6
MnO	42.30	0.03	42.3
FeO *	4.34	n.a.	4.3

).

3. Results

3.1. PXRD

In all three samples, wollastonite (CaSiO₃; ICDD PDF2: 01-084-0654), as well as the Li-manganates LiMnO₂ (ICDD PDF2: 00-035-0749) and Li₂MnO₃ (ICDD PDF2: 01-081-1953) could be detected with PXRD. Rankinite (Ca₃Si₂O₇; ICDD PDF2: 01-076-0623) was detected in slags 1 and 2 and larnite (Ca₂SiO₄; ICDD PDF2: 00-029-0369) in slag sample 3. The presence of hausmannite (Mn²⁺Mn³⁺₂O₄; ICDD PDF2 00-024-0734) and the Li-silicate Li₂SiO₃ (ICDD PDF2: 00-029-0828) is conceivable. However, the presence of these two phases cannot be clearly verified with this method due to line overlaps. Schnickmann et al. [8] have already discovered that within the Li-manganate LiMnO₂, the elements Li and Mn can replace each other, which can result in shifting to smaller or larger lattice parameters. The same applies to the replacement of Mn²⁺ in Ca-silicate matrices. An exchange of elements in the crystal lattice is only possible as long as the exchangeable elements have approximately the same ionic radii (e.g., [31]). An overview of the recorded diffractograms of the three slags is given in Figure 5.

3.2. EPMA

With EPMA, the following phases were determined:

• Li_(1−x)Mn_(1+0.33x)O₂ / Li_(1+x)Mn_(1−0.33x)O₂ (slags 1, 2, 3)
• Li_(2−x)Mn_(1+0.33x)O₃ / Li_(2+x)Mn_(1−0.33x)O₃ (slag 2)
• Li₂SiO₃ (slags 1, 2)
• Li_xMn_(3−x)O₄ (slags 1, 2, 3)
• CaSiO₃ (slags 1, 2, 3)
• Ca₃Si₂O₇ (slags 1, 2)
• Ca₂SiO₄ (slag 3)
• Residual melt (slags 2, 3)

These phases were identified based on the elemental content of the individual grains (point measurements, line scans) and the crystal shape. The crystal shape provides additional information regarding the processes during solidification (Figure 6).

3.2.1. Li-Manganates

Two different Li-manganates (Li₂Mn⁴⁺O₃ and LiMn³⁺O₂) were found in the slag samples. Li₂MnO₃ was only detected via EPMA in slag 2 in a prepared, near-surface region of the slag where 2 wt.% Mg was added. This phase theoretically contains 47.03 wt.% Mn and 11.88 wt.% Li. Within this phase, the average measured Mn content was 47.08 wt.% (min.: 46.88 wt.%; max.: 47.28 wt.%) (Figure 7). In addition, this phase forms idiomorphic crystals between 30 and 100 µm. Based on the crystal shape, it can be determined that Li₂MnO₃ forms early during solidification. Point measurements and recorded line scans within this phase show that no other elements (e.g., Mg, Ca, Si) have been incorporated. According to the line scan results, the Mn content within the crystal is constant (Figure 8).

The crystallized LiMnO₂ formed idiomorphic to partially hypidiomorphic crystals with an average size of 50 µm. In addition, elongated LiMnO₂ needles crystallized in the Ca-silicate matrix. In theory, this phase contains 58.52 wt.% Mn and 7.39 wt.% Li. Within the slag, the average Mn content in S1 was 58.52 wt.% (min.: 58.25 wt.%; max.: 58.76 wt.%), in S2 it was 58.67 wt.% (min.: 58.35 wt.%; max.: 58.88 wt.%) and in S3 it was 57.88 wt.% (min.: 57.02 wt.%; max.: 58.17 wt.%) (Figure 8). Slag 3 contains 0.83 wt.% Al, which results in a lower Mn concentration (

Table 3. For the two Li-manganates LiMnO₂ and Li₂MnO₃, the average structural formula was calculated from all measuring points. To clarify the possible exchange of Li and Mn ions in the crystal lattice, a structural formula was also calculated for the lowest and highest Mn concentration according to: Li_(1−x)Mn_(1+0.33x)O₂/Li_(1+x)Mn_(1−0.33x)O₂ and Li_(2−x)Mn_(1+0.33x)O₃ / Li_(2+x)Mn_(1−0.33x)O₃.

Sample	LiMnO2 (Average)	LiMnO2 (min.)	LiMnO2 (max.)
S1 (n = 25)	Li1.01+Mn3+1.0O2	Li1+1.01Mn3+0.99O2	Li1+0.98Mn3+1.01O2
S2 (n = 25)	Li1+1.0Mn3+1.0O2	Li1+1.01Mn3+1.00O2	Li1+0.97Mn3+1.01O2
S3 (n = 25)	Li1+1.0(Mn3+0.97Al3+0.03)O2	Li1+1.02(Mn3+0.96Al3+0.03)O2	Li1+0.96(Mn3+0.99Al3+0.02)O2
Sample	Li2MnO3 (average)	Li2MnO3 (min.)	Li2MnO3 (max.)
S2 (n = 25)	Li1+2.0Mn3+1.00O3	Li1+2.02Mn3+0.99O3	Li1+1.97Mn3+1.01O2

). The results of the line scans show that the Mn content within the individual grains is almost homogeneous. In addition, no decrease in element concentration is observed towards the grain boundaries (Figure 9).

3.2.2. Hausmannite and Spinel

Lithium-rich hausmannite was found in all three slag samples. Hausmannite consists of a mixture of Mn²⁺ and Mn³⁺ occupying tetrahedral and octahedral sites, respectively. When Mn²⁺ is substituted by Li¹⁺, Mn⁴⁺ can appear in the octahedral sublattice to balance the charge without a change in oxygen stoichiometry. However, there is not enough crystallographic data for the site occupancies of delithiated hausmannite. Additionally, tetragonal Mn₃O₄ and orthorhombic LiMnO₂ are intergrown. With an average crystal size of 10–30 µm, these crystals are significantly smaller than LiMnO₂. In all three slag samples, the incorporation of other elements into the crystal lattice was detected. The incorporation of Li was recorded in all slag samples. The largest amount of Li was incorporated into S1, with 1.59 wt.%, followed by S2 with 0.80 wt.% and S3 with 0.51 wt.%. In addition, an average incorporation of 0.84 wt.% Mg and 0.82 wt.% Al was observed in S3. For the calculation of the spinel solid solution in S3, the following virtual compounds were defined: Mn₃O₄, MnAl₂O₄, MgAl₂O₄ and Li₂Mn₂O₄. This solid solution can be expressed by the following generalized stoichiometric formula: (Li_(2x),Mg_(x),Mn_(1−2x))_1+x(Al_(2−z),Mn³⁺_(z))₂O₄. For Li-rich hausmannite in S1 and S2, the following generalized stoichiometric formula was applied: (Li_(2x)Mn_(1−x))_1+x(Mn³⁺)₂O₄ (

Table 4. Determined structural formula of lithiated hausmannite in slags 1,2 and 3.

Sample	Mn2+Mn3+2O4
S1 (n = 12)	(Li1+0.50Mn2+0.75)Mn3+2.0O4
S2 (n = 16)	(Li1+0.26Mn2+0.87)Mn3+2.0O4
S3 (n = 16)	(Li1+0.16Mn2+0.84Mg2+0.08)(Mn3+1.93Al3+0.07)O4

).

3.2.3. Li-Silicate

The Li-silicate Li₂SiO₃ was only found in S1 and S2. This phase forms 20 to 50 µm, rounded, elongated- to needle-shaped crystals. The hypidiomorphic to xenomorphic crystal shape indicates crystallization towards the end of solidification. The ideal crystals of this phase contain an average Si content of 31.22 wt.% and an average Li content of 15.43 wt.%. In S1, the average Si content is 31.66 wt.% and in S2 it is 31.68 wt.%. A higher Si content indicates a lower Li content.

3.2.4. Matrix-Forming Phases

The main matrix-forming mineral is wollastonite (CaSiO₃), followed by rankinite (Ca₃Si₂O₇) in S1 and S2 and larnite (Ca₂SiO₄) in S3. All three phases form xenomorphic crystals, which indicates crystallization at the end of solidification. Due to the approximately equal ionic radii of Mg²⁺, Ca²⁺ and Mn²⁺, Ca can be replaced by Mg and Mn in the crystal lattice [31]. In S1, 0.85 wt.% Mn was incorporated in wollastonite and 1.28 wt.% was incorporated in rankinite; in S2, there was 0.59 wt.% Mn in wollastonite and 2.22 wt.% in rankinite; in S3, there was 0.75 wt.% Mn in wollastonite and 6.97 wt.% Mn and 0.76 wt.% Mg in larnite (

Table 5. Calculated structural formulas for the matrix-forming phases wollastonite (CaSiO₃) and rankinite (Ca₃Si₂O₇) in S1 and S2 and of larnite (Ca₂SiO₄) in S3.

Sample	CaSiO3
S1 (n = 40)	(Ca2+0.99Mn2+0.02)Si4+0.99O3
S2 (n = 35)	(Ca2+0.99Mn2+0.01)Si4+1.0O3
S3 (n = 30)	(Ca2+0.99Mn2+0.02)Si4+1.0O3
Sample	Ca3Si2O7
S1 (n = 40)	(Ca2+2.96Mn2+0.07)Si4+1.99O7
S2 (n = 40)	(Ca2+2.86Mn2+0.12Mg2+0.02)Si4+2.0O7
Sample	Ca2SiO4
S3 (n = 28)	(Li1+0.33Ca2+1.63Mn2+0.14Mg2+0.05)Si4+0.99O4

).

4. Discussion

The aim of this study was to investigate the stabilization of Mn⁴⁺ in the slag and to also verify which Li-rich phases are formed. Moreover, it should be investigated whether Mn⁴⁺ forms a compound with Mg or Al. Based on the phase diagrams, the experiment provides an overview of which phases crystallize in a synthetic slag consisting of Li₂O, MgO, Al₂O₃, SiO₂, CaO and MnO_x.

4.1. Influence of Mg and Al on Mn⁴⁺

As can be seen from Figure 3e, the spinel phase in the MgO-Al₂O₃ system is the stable phase from low temperatures and has an extension towards both Al₂O₃ and MgO concentrations at high temperatures. The addition of MgO together with Al₂O₃ will increase the stability range of cubic spinel in the Li₂O-MnO_x-MgO-Al₂O₃ system.

The tetragonal-to-cubic transformations for the spinel phase in the MgO-MnO_x and Al₂O₃-MnO_x systems are driven by the Jahn–Teller distortion of octahedral sites, primarily occupied by Mn³⁺ ions. In the cubic phase, assuming a disproportionation of Mn³⁺ into Mn⁴⁺ and Mn²⁺ and the occupation the octahedral sites by Mn⁴⁺ and distribution of Mn²⁺ between tetrahedral and octahedral sites, a general formula for cubic spinel (Al³⁺, Mn²⁺)₁(Al³⁺, Mn³⁺, Mn²⁺, Mn⁴⁺, Va)₂(Mn²⁺, Va)₂(O²⁻)₄ is employed in the Al₂O₃-MnO_x system and (Mg²⁺, Mn²⁺)₁(Mg²⁺, Mn³⁺, Mn²⁺, Mn⁴⁺,Va)₂(Mg²⁺, Mn²⁺, Va)₂(O²⁻)₄ in the MgO-MnO_x system. This formulation accommodates the variation in oxidation states of Mn within the crystal lattice. The cubic spinel phase is further characterized by the inclusion of vacancies (Va) on the octahedral sites and Mn²⁺/Mg²⁺ in the interstitial site. The vacancies are introduced into the spinel model to extend the homogeneity ranges towards the Al₂O₃ side and interstitial site to model extension towards the MnO_x/MgO side. The applied models account for deviations from a perfectly ordered structure to describe the disordering in cationic sites and to describe the homogeneity ranges of spinel, thus enhancing the model applicability in a large range of conditions (compositions, temperature, oxygen partial pressure). The modeling of tetragonal spinel differs from that for cubic spinel by introducing Mn³⁺ into the tetrahedral site and via the absence of Mn⁴⁺ in the octahedral site in both systems. The tetragonal phase is stable at lower temperatures and in a more narrow composition range than cubic spinel.

The octahedral sites are substantially occupied by Mn⁴⁺ in cubic spinel in the MgO-MnO_x system in the whole possible composition range. A low concentration of Mn⁴⁺ is found in the spinel phase, with compositions close to stoichiometric MnAl₂O₄ in the Al₂O₃-MnO_x system even at high oxygen partial pressures. Thus, the introduction of Mg into the spinel phase not only enhances the stability but also effectively increases the Mn⁴⁺ concentration. This suggests a more pronounced role for Mg in maintaining the desired Mn oxidation state within the spinel structure.

The Li addition introduces a notable impact on the Mn⁴⁺ ion concentration on the octahedral sites. Substitution of Mn²⁺ for Li¹⁺ in the tetrahedral sites reduces the charge of cations in spinel. In order to preserve the electroneutrality of this structure, the fraction of Mn⁴⁺ in the octahedral sites occupied by Mn³⁺ in tetragonal spinel should be increased. The increase in Mn⁴⁺ in octahedral sites should also reduce the Jahn–Teller effect caused by the electronic structure of the Mn³⁺ cation. It should be noted that in the stoichiometric spinel LiMn₂O₄, the cation Li¹⁺ completely occupies the tetrahedral site and the ratio of Mn⁴⁺/Mn³⁺ in octahedral site is equal to one. In cases where the Li content is higher than in stoichiometric spinel, Li¹⁺ partially occupies octahedral site and ratio Mn⁴⁺/Mn³⁺ becomes larger than one. With a temperature increase, the spinel becomes enriched by Mn, with a shift of composition to the stoichiometric spinel LiMn₂O₄ and an even higher content of Mn. Therefore, spinel with a high Li content is stable at lower temperatures and with a temperature increase it will decompose to spinel with a lower Li content and a Li₂MnO₃ phase (Figure 2). At a temperature of 1230 K, LiMnO₂ forms from spinel and Li₂MnO₃ and, at temperatures of 1250 K, Li₂MnO₃ decomposes into LiMnO₂ and Li₂O. Tetragonal spinel with a composition close to the stoichiometric one is stable in equilibrium with hausmannite with a much lower Li content from one side and from other side it is in equilibrium with LiMnO₂. The high-temperature stability limit of tetragonal spinel is defined by decomposition to hausmannite and LiMnO₂ at 1270 K.

According to the phase diagram shown in Figure 3b, no reactions should take place between MnO₂ and Al₂O₃ up to 688 K. The experimental results indicate that Mg and Al react to form spinel (Mg₂AlO₄) and therefore small amounts of these elements are not a problem for Li-manganate (LiMnO₂ or Li₂MnO₃) forming processes. As long as Mn²⁺ and Mn³⁺ are present in the slag, spinels with Mg, Al and Mn are formed, which can interfere with the formation of EnAMs. In order to produce large and pure EnAM crystals that can also be efficiently separated from the rest of the slag, it must be ensured that LiMnO₂ or Li₂MnO₃ crystallize out of the melt first.

4.2. Li-Rich Phases as a New Potential EnAM

A total of four Li-containing phases crystallized out in the slag. These phases include the two Li-manganates LiMnO₂ and Li₂MnO₃, as well as the Li-silicate Li₂SiO₃ and the Li-containing hausmannite. With an average of 0.51 wt.% (S3) to 1.59 wt.% Li (S1), the Li content in hausmannite is too low to be a potential EnAM. The Li¹⁺ is incorporated at the tetrahedral site and replaces Mn²⁺ in the crystal lattice. The incorporation of Li¹⁺ into cubic LiMn₂O₄ can occur by balancing Mn⁴⁺ on the octahedral site, but it can also occur by deviation from the stoichiometry of oxygen by introducing vacancies [32,33,34]. There is not enough crystallographic data for hausmannite showing how substitution of Mn²⁺ by Li¹⁺ is compensated. Since it is not possible to detect small amounts of Mn⁴⁺ with EPMA, more advance techniques should be used to further investigate whether Mn⁴⁺ can be incorporated into the crystal lattice of hausmannite for charge compensation. Hausmannite forms in addition to relatively tiny crystals and only occurs in small amounts. Due to the partially inhomogeneous mixture of Mn²⁺ and Mn³⁺ in this phase, further processing would be necessary after successful separation from the slag. With an average of 15.43 wt.% Li, Li₂SiO₃ would be more suitable as a new EnAM. However, this phase crystallizes towards the end of solidification and thus forms relatively small and needle-shaped crystals that are inhibitors for a successful separation of Li. Furthermore, Li and Si must be separated from each other after the recycling step. Therefore, Li-manganates like LiMnO₂ or Li₂MnO₃ would be more suitable as new potential EnAMs. The Li-manganate Li₂MnO₃, with 11.88 wt.% Li, could only be found in slag 2 where 2 wt.% Mg was added. No incorporation of other elements could be observed within this phase. As shown in Figure 2, the Li₂MnO₃ phase is stoichiometric, which corroborates with the absence of the solubility of other elements. Furthermore, the Li₂MnO₃ phase has some advantages as a new EnAM. Firstly, it is an early crystallite with large, idiomorphic crystals that contains more Li than LiAlO₂ (10.35 wt.% Li). The stabilization of only Mn⁴⁺ in the slag has the additional advantage that spinel formation with Mn could be suppressed and that the Jahn–Teller effect does not cause deformation. In addition, LiMnO₂ (7.39 wt.% Li) can also be considered as a potential EnAM. The low incorporation of Al (0.83 wt.%) does not seem to have any effect on the crystal lattice. According to Kong et al. [14], small amounts of Al in the Li-rich cathode material should promote intralayer diffusion.

Nevertheless, it remains to be determined which of the two Li manganates, LiMnO₂ or Li₂MnO₃, formed first. It is possible that LiMnO₂ forms first as a high-temperature phase and remains stable during cooling. At low temperatures, according to phase diagram, the stable phases are spinel and Li₂MnO₃. The alternative would be that Li₂MnO₃ forms first and decomposes to LiMnO₂ and Li₂O, which reacts with SiO₂ forming Li₂SiO₃.

5. Conclusions and Outlook

The presented results indicate that it was possible to stabilize the Mn⁴⁺-containing Li-manganate Li₂MnO₃ in a synthetic slag, next to LiMnO₂. Both Li-manganates, Li₂MnO₃ and LiMnO₂, have proven to be extremely pure phases. Using EPMA for phase identification, LiMnO₂ was found in all areas of the prepared slag, while Li₂MnO₃ was found in only one area, the near-surface area of the slag that was affected by oxygen. Accordingly, based on the literature data for other slag systems, it is assumed that the added oxygen only affects the near-surface region due to the low viscosity of the melt [35,36]. In addition, it needs to be clarified which elements can be added to the system to reduce the viscosity. For this purpose, future experiments should investigate whether, as described in the literature (e.g., [37]), the viscosity can be reduced by adding FeO and Fe₂O₃ so that the oxygen can affect the entire melt, with the aim of stabilizing Mn⁴⁺ throughout the slag. Stabilization of Mn⁴⁺ could provide further important experimental data on whether spinel formation can be prevented or whether Mn⁴⁺ from compounds with elements other than Li.

However, the conditions under which Li₂MnO₃ was formed and how Mn⁴⁺ was stabilized throughout the slag need to be clarified. A promising approach would be to investigate phase relations (experimentally and theoretically) and to develop thermodynamic databases for a complex system in order to understand the processes taking place during solidification. Further investigations should indicate which slag modification should be performed to maximize EnAM yield.