Formation of Lithium-Manganates in a Complex Slag System Consisting of Li₂O-MgO-Al₂O₃-SiO₂-CaO-MnO—A First Survey

Abstract

Due to the increasing demand for electromobility, the recovery of technologically relevant elements from spent Li-ion batteries is becoming increasingly important. Pyrometallurgical processing can deal with a broad range of input materials. Unfortunately, ignoble elements such as Li and Mn enter the slag. A novel approach to facilitate this processing is the Engineered Artificial Minerals (EnAM) strategy for the recovery of critical elements. The aim of this study is to investigate whether it is possible to stabilize Li in Li-manganates as the first crystallizate. For this purpose, synthetic oxide slags (Li, Mg, Al, Si, Ca, Mn) of varying compositions were made. The constituting compounds were identified using inductively coupled plasma optical emission spectrometry, powder X-ray diffraction, X-ray absorption near-edge structure analysis, and electron probe microanalysis. These results provide an understanding of the solidification process and the behavior of the elements of concern. Lithium-manganate(III) (LiMnO₂) crystallized first, next to hausmannite (Mn²⁺Mn³⁺₂O₄) in a matrix consisting of wollastonite (CaSiO₃) and larnite (Ca₂SiO₄). Within the structure of LiMnO₂, Li and Mn can replace each other in certain proportions. By adding Al and Mg spinel, solid solutions between Mn²⁺Mn³⁺₂O₄, MnAl₂O₄, MgAl₂O₄ and LiMnO₂ are expected and described by the stoichiometry formula: (Li_(2x),Mg_(1x),Mn_(2+(1–x)))_1+x(Al_(2–z),Mn³⁺_(z))₂O₄.

1. Introduction

At present, modern applications in electrics and electronics are predominantly based on the use of Li-ion batteries (LIBs). These batteries are mainly utilized in electric vehicles and storage systems, but also in laptops or mobile phones, and have therefore different lifetimes [1,2]. Conversely, this means that more and more natural resources are needed, including technologically relevant elements such as Li. The global reserves of Li are not evenly distributed and are not readily available in every country [3]. According to the United States Geological Survey [4], Li production increased by 21% from 2021 to 2022 (global reserves: 26,000,000 t). Mining of this element is limited to certain countries. Australia controls nearly 50% of mine production, followed by China and Chile [3,4,5]. The recycling of end-of-life LIBs is important in order to reduce dependence on these countries, close material cycles (keyword: circular economy) and avoid future raw material shortages. Common recycling methods for LIBs include the hydrometallurgy and pyrometallurgy (or a combination of both) routes [5,6,7,8]. During the pyrometallurgical recycling process, Li, as well as other elements such as Al or Mn, pass into the slag as an ignoble element or are lost as dust [1,9,10]. In order to recover the Li efficiently, cost-effectively and economically from the slag (e.g., by flotation), the element should be enriched in a single phase (Figure 1). That is the idea behind the Engineered Artificial Minerals (EnAM) approach. The procedure to effectively implement the EnAM strategy is based on characterizing the slag, identifying potential EnAMs and then modifying the slag system to optimize the amount, crystal size and habitus of this target phase. In unmodified industrial slags, Li is usually present in several different phases as well as partly in the matrix [10,11].

Elwert et al. [12] suggest that Li-aluminate (LiAlO₂) is a promising EnAM for Li recovery because this compound occurs quite early during solidification and forms large idiomorphic to hypidiomorphic crystals [11,12]. These idiomorphic to hypidiomorphic crystals can be isolated from the rest of the slag via flotation [13]. During solidification and crystal formation, chemical processes occur that affect the formation of LiAlO₂. For instance, besides LiAlO₂, spinels are also formed as early crystals (XY₂O₄ with a cubic crystal structure), scavenging the Al out of the melt. This reaction occurs mainly when elements such as Mn or Mg are present in the slag system [14,15].

In LIBs, Li is used in the anode as Li_xC_y (x = 1 and y = 6, typically) and in the cathode, e.g., as lithium metal oxides (LiMO₂), commonly LiCoO₂ (LCO) [7,16,17]. Due to the lower cost and the higher theoretical specific capacity, LiNi_xMn_yCo_zO₂ (x + y + z = 1; NMC) layered oxides will prevail over LCO [1,6,18]. In future slag systems, Mn will be present due to the use of NMC, presumably as a more cost-effective and health-friendly transition metal oxide in novel cathode materials. The spinel LiMn₂O₄ (LMO), LiMnO₂ or Li and Mn-rich layered oxides have been intensively studied for Co replacements [16,19,20,21,22]. Despite the Jahn-Teller deformation of the Mn³⁺, which can destabilize the lattice framework, ongoing studies indicate that pure LiMnO₂ electrodes will be available in the future [23]. When using pyrometallurgical processing to regain noble elements such as Co, Ni and Cu, the Li and Mn from LiNi_xMn_yCo_zO₂ will be present in future slags along with Al (cell and module cases) [24,25]. Thus, spinels (MgMn₂O₄, MgAl₂O₄, MnAl₂O₄ and Mn²⁺Mn³⁺₂O₄) will form as early crystallizates [26,27]. Spinel formation will decrease the amount of the EnAM LiAlO₂ [10,11]. In addition to the competition from spinel, this phase (LiAlO₂) also has a (limited) solid solution with SiO₂, resulting in a Si concentration of 2–3 wt.% Si [13]. This would pose a challenge for subsequent hydrometallurgical processing (e.g., the need for hydrofluoric acid) for the recovery of Li.

For this reason, it was investigated whether a Li-manganate could serve as a suitable EnAM. In this way, Li and Mn could be recovered simultaneously. In an optimal outcome, the recovered Li-manganate (e.g., as LiMnO₂) could be directly used again as cathode material.

Therefore, it is investigated whether Li-manganates can be stabilized in the slag system. Parameters potentially influencing the type of Mn compounds formed during solidification, such as the Mn concentration and oxidation state, need to be considered [28].

For this purpose, three different lab-scale melting experiments were performed under normal atmospheric conditions to obtain synthetic slags. The system of all experiments included Li and Mn, as well as Ca and Si, as slag-forming compounds. In one experiment, 2 wt.% Mg and in another, 2 wt.% Mg and 2 wt.% Al were added. The bulk qualitative mineralogical composition was determined by powder X-ray diffraction (PXRD). Electron probe microanalysis (EPMA) was used to analyze the element concentration of single grains. Lithium could not be determined with the EPMA system; therefore, virtual compounds were used to calculate the concentrations [29]. The use of virtual components containing a readily measurable element such as Mn enable better results than the direct measurement of, e.g., oxygen with limited precision. Therefore, for example, the stoichiometric calculation of oxygen is common practice in microprobe analysis. To determine the correct Li-manganates in the system, it is necessary to accurately determine the Mn concentration on the one hand and the oxidation state on the other. The Mn speciation was determined with laboratory-based X-ray absorption near edge structure (XANES).

2. Background

Here, the most important information about binary and ternary systems of the components Li₂O, MgO, Al₂O₃, SiO₂, CaO and MnO are summarized, with a special focus on the Mn-oxides and Li-manganates. They will enable a better understanding of the discussed reactions and processes.

2.1. Important Binary Phase Systems for Li-Manganates and Mn-Oxides

Due to the large number of Li-Mn-O compounds (

Table 1. List of known Li-Mn-oxides. (1): taken from the materials project (preference); (2): Freire et al. (data from Ref. [30]); (3): Levi et al. (data from Ref. [31]). Red-marked rows: Important for this study. Indication of Mn and Li concentration in wt.%.

Formula	Common/Observed System	Mn2+	Mn3+	Mn4+	Mntotal	Li	Described in (Example)
MnO/m3m	Fm3̅m (1)	77.45			77.45	0.00	e.g., Slagatlas
Li2Mn3O4	-	67.91			67.91	5.72	Uncommon
Li2Mn2O3	P12₁/c (1)	63.97			63.97	8.08	Mater. P. mp-756298
Li2MnO2	P3̅m1 (1)	54.49			54.49	13.77	Mater. P. mp-19279
Mn5O6	-	44.46	29.64		74.10	0.00	Wei et al. [20]
Mn4O5	-	36.66	36.66		73.32	0.00	Uncommon
Mn3O4	spinel (1)	24.01	48.02		72.03	0.00	e.g., Slagatlas
LiMn3O4	C12/m1 (1)	46.60	23.30		69.90	2.94	Xu et al. [21]
LiMn2O3	C12/c1 (1)	33.30	33.30		66.60	4.21	Uncommon
Li3Mn2O5	P1̅, C12/m1 (1)	28.22	28.22		56.44	10.70	Uncommon
Mn2O3	Pbca (1)		69.60		69.60	0.00	e.g., Slagatlas
LiMn5O8	spinel (1)		67.06		67.06	1.69	Uncommon
LiMnO2	Pmmn (1)		58.52		58.52	7.39	M. M. Thackeray [22]
Li4Mn2O5	Fm/3m (2)		50.49		50.49	12.76	Freire et al. [18]
Li3MnO3	P12₁/c1 (1)		44.39		44.39	16.83	Uncommon
Li8Mn2O7	P12₁/c1 (1)		39.61		39.61	20.02	Uncommon
Li5MnO4	Pbca (1)		35.76		35.76	22.59	Hoang et al. [23]
Li12Mn2O9	P1 (1)		32.59		32.59	24.70	Uncommon
LiMn6O12	P1 (1)		10.39	51.97	62.36	1.31	Uncommon
LiMn5O10	P1 (1)		12.44	49.76	62.20	1.57	Uncommon
LiMn4O8	P2_13̅ (1)		15.49	46.47	61.96	1.96	Uncommon
Li0.3MnO2	spinel (3)		18.51	43.20	61.71	2.34	Levi et al. [19]
LiMn3O6	C1c1 (1)		20.52	41.04	61.56	2.59	Uncommon
LiMn2O4	Fd3̅m1 (1)		30.38	30.38	60.76	3.84	Mishra and Ceder [24]
MnO2	P4₂/mnm (1)			63.19	63.19	0.00	e.g., Slagatlas
Li2Mn4O9	P3̅c1 (1)			58.19	58.19	3.68	Masquelier et al. [25]
Li4Mn5O12	C12/c1 (1)			55.56	55.56	5.62	Masquelier et al. [25]
Li2MnO3	C12/m1 (1)			47.03	47.03	11.88	Wang et al. [26]
Li4MnO4				37.45	37.45	18.93	Uncommon
Li1+xMn2–xO4		0.00–30.38	30.38–55.56	55.56–60.77	3.84–5.62		Paulsen and Dahn [27]
Li1-xMnO4			0.00–30.38	30.38–63.19	60.77–63.19	0.00–3.84	Calderon et al. [28]
LixMn3–yO4		24.01–53.77	15.76–48.02		69.24–72.03	0.00–3.88	Paulsen and Dahn [27]

), the special focus of this work is mainly on the compounds discovered/important in the samples. The relevant systems are discussed based on selected key publications.

For this study, the system (Li₂O-MnO_x, 1 ≤ x ≤ 2) is one of the most important binary systems. Several Li-manganates are known in the literature, with different concentrations and Mn oxidation states (Table 1). Some phases are described in more detail by Longo et al. [32] and Paulsen and Dahn [33]. Paulsen and Dahn [33] focus on the Li-spinel (Li_(1+x)Mn_(2–x)O₄), which has a high stability field between 400 and 880 °C. The structure ranges from LiMn₂O₄ (Mn^3.5+) to Li₄Mn₅O₁₂ and can coexist with Li₂MnO₃ and/or Mn₂O₃. At temperatures above 880 °C, both Li₂MnO₃ and LiMn₂O₄ decompose into layered LiMnO₂, Mn₃O₄ and O₂. Below 400 °C, the only stable spinel compounds are LiMn_1.75O₄ as well as Li₄Mn₅O₁₂. The spinels LiMn₂O₄ and Li₂Mn₃O₇ are thermodynamically stable under acidic conditions. (Refs. [32,33,34]. In a temperature range between 1000 and 1060 °C, lithiated hausmannite (Li_xMn_3–yO₄, 0 ≤ x ≤ 1.33) coexists with LiMnO₂ (rock salt type, orthorhombic). [33,34]. In the literature, no information is found on whether Mn²⁺ or Mn³⁺ is replaced by Li in lithiated hausmannite. In terms of ionic radii, Mn²⁺ would have to be replaced by Li [35]. In addition to temperature and pressure, the pH value also has an influence on the stability range of the Li-manganates. In addition, Mishra and Ceder [36] describe which compounds in the Li₂O-MnO_x system are affected by the Jahn-Teller distortion. According to the authors, layered structures (LiMnO₂) and cubic spinels (LiMn₂O₄) are not affected. However, monoclinic layered and orthorhombic structures (e.g., LiMnO₂) as well as tetragonal spinels (LiMn₂O₄) are affected [36].

An Important Li-free spinel in this system is hausmannite Mn²⁺Mn³⁺₂O₄ (MnO-Mn₂O₃). Under normal conditions (T = 25 °C; 1 atm), hausmannite can be considered a deformed spinel with a tetragonal crystal lattice. This deformation is due to the Jahn-Teller effect below 1172 °C from a deformation of the coordination octahedron around the Mn³⁺ position (M 2) [15,37,38].

The systems MnO-Al₂O₃ and Li₂O-Al₂O₃ play a rather subordinate role for consideration in this work since these occur only as virtual components in the calculation of the Li-manganates.

One of the most relevant minerals in the MnO-Al₂O₃ system is galaxite (MnAl₂O₄). This mineral is part of a solid solution series in the cubic spinel (XY₂O₄) series. The Y position (octahedral site) is occupied mainly by Al and, above 1172 °C, also by Mn³⁺. Whereas the X position (tetrahedral site) is mainly occupied by Mn or Mg. In the cubic form of Mn₃O₄, gapless solid solutions with MnAl₂O₄ as well as MgAl₂O₄ are possible. For the tetragonal form, only limited mixtures are possible [15,39].

Important stable phases in the Li₂O-Al₂O₃ system described by Konar [39] are: Li₅AlO₄, LiAlO₂, LiAl₅O₈ (spinel), and LiAl₁₁O₁₇. For LiAlO₂, the most known polymorphs are the α-phase (low temperature, cubic) and the γ-phase (high temperature, tetragonal).

2.2. Important Ternary Phase Systems Containing Li and Mn

A system consisting of Li₂O, Al₂O₃ and MnO_x has not yet been described in the literature.

In the Li₂O-MgO-Al₂O₃ system, spinel solid solutions between MgAl₂O₄ and LiAl₅O₈ are described. The resulting mixture can be expressed by the following formula: (Li⁺, Mg²⁺, Al³⁺)_{[tetrahedral site]}(Li⁺, Mg²⁺, Al³⁺, Va)_{2[octahedral site]}O₄ [31]. There also is a solid solution between MgAl₂O₄ and γ-Al₂O₃ [11,12].

In the MgO-MnO-Al₂O₃ system, observed by Park et al. [40], spinel solid solutions between MgAl₂O₄ and MnAl₂O₄ are described.

Stable occurring compounds in the Li₂O-Al₂O₃-SiO₂ (LAS) system include eucryptite (LiAlSiO₄), spodumene (LiAlSi₂O₆) or petalite (LiAlSi₄O₁₀). Further (meta)stable phases in the system are described by Konar et al. [41]. Because of the limited solid solution with Si, LiAlO₂(Si) can be seen as part of this system [41].

Konar et al. [42] describe stable phases in the Li₂O-MgO-SiO₂ system. In the orthosilicate section (Li₄SiO₄-Mg₂SiO₄), Li₂MgSiO₄ can be included considering the miscibility gap between Li₄SiO₄-Li₂MgSiO₄. In the Li₄SiO₄ system, Li is replaced by Mg at the large-sized M-site (4e site).

In the CaO-SiO₂-MnO system, solid solutions between Ca₂SiO₄, Mn₂SiO₄ and Mg₂SiO₄ are possible, as well as between CaSiO₃, MnSiO₃ and MgSiO₃. The minerals larnite (Ca₂SiO₄) and wollastonite (CaSiO₃) are important matrix-forming phases occurring within various slag systems. Larnite belongs to the olivine-group and wollastonite belongs to the pyroxene-like group. Under reducing conditions, both phases can incorporate Mn²⁺ and/or Mg²⁺ in the Ca position [43,44]. The ionic radius of Ca is slightly larger than that of Mn or Mg; hence, Ca occupies the larger M₂-site [35,45].

3. Materials and Methods

3.1. Materials

The precursors for the melting experiments were prepared in the laboratory using the gel-combustion method. The chemicals used for the experiments were: Lithium nitrate (99.5%, Roth, Karlsruhe, Germany), magnesium nitrate hexahydrate (98%, Carl Roth, Karlsruhe, Germany), aluminum nitrate nonahydrate (98%, Carl Roth, Karlsruhe, Germany), Köstrosol^® 0830 AS (Chemiewerk Bad Köstritz GmbH, Bad Köstritz, Germany), calcium nitrate tetrahydrate (99%, Carl Roth, Karlsruhe, Germany), manganese(II) nitrate tetrahydrate (98%, Carl Roth, Karlsruhe, Germany), and citric acid (99.5%, VWR chemicals, Darmstadt, Germany). The mass fractions for the required products used for the three experiments can be taken from

Table 2. Element concentrations used for the preparation of the precursors, given in weight percent (wt.%). In S2, 2 wt.% of Mg and in S3, 2 wt.% of Mg and Al each were added. In all three experiments, the ratios used were Ca:Si = 1.01 and Mn:Li = 2.90.

Sample	S1	S2	S3
Li2O	4.1	4.1	3.8
MgO	n.a.	2.0	2.0
Al2O3	n.a.	n.a.	2.0
SiO2	41.3	40.8	40.4
CaO	41.8	41.2	40.6
MnO2	12.7	11.9	11.2

.

For the preparation of Li-manganate(IV) (Li₂MnO₃) lithium hydroxide (98%+, Merck AG, Darmstadt, Germany) and citric acid monohydrate (99.5%, for analysis, Merck AG, Darmstadt, Germany) were used and for the preparation of galaxite (MnAl₂O₄) aluminum nitrate nonahydrate (Carl Roth) and manganese(II) nitrate tetrahydrate (Carl Roth).

The used references for XANES include: Mn²⁺O (alfa aesar, Thermo fisher GmbH, Kandel, Germany, 99%), Mn^2.67+O₄ (Sigma-Aldrich, St. Louis, MI, USA, 97%), LiMn³⁺O₂ (Sigma-Aldrich, St. Louis, MI, USA, 99%), Mn³⁺₂O₃ (Sigma-Aldrich, St. Louis, MI, USA, 99%) and Mn⁴⁺O₂ (Merck AG, Darmstadt, Germany, for synthesis, 86.94%). Microcrystalline cellulose Avicel^® (Merck AG, Darmstadt, Germany) was used for the pellet preparation.

3.2. Methods

3.2.1. Experimental

The gel-combustion method was used to prepare a homogeneous precursor for the melt experiments. Common procedures [46,47] were adapted accordingly for the synthesis. The nitrates were dissolved in deionized water, and then the citric acid was added. Next, the solution was placed on a heating plate and slowly reduced from 200 to 250 °C. After the solution had a yellowish color, the SiO₂ nanoparticles were added to the solution and the heating plate was set to 400 °C for self-combustion. The resulting pre-precursor was oxidized with ammonium nitrate in quartz crucibles at a heating rate of 10 °K/min to 480 °C in a muffle furnace (Nabertherm LE 1/11/R7, Nabertherm GmbH, Lilienthal, Germany) to remove the remaining carbon.

For the preparation of the reference material, Li₂MnO₃, the gel-combustion method was also used. However, the pre-precursor was thermally treated (1000 °C; 3 h) in air in a muffle furnace. The used galaxite (MnAl₂O₄) was prepared according to Wittkowski et al. [9] via solution combustion synthesis.

3.2.2. Melt Experiment

The treated precursors were melted under a normal atmosphere in Pt/Rh crucibles in a chamber furnace (Nabertherm HT16/17, Nabertherm GmbH, Lilienthal, Germany). For the melting process, the program from Schirmer et al. [13] and Wittkowski et al. [11] was adapted (Figure 2).

3.2.3. Determination of Melting Points

For an adjusted melting program, the melting point of the thermally treated precursors was determined via a heating microscope (EM301, Hesse Instrument e. K., Osterode am Harz, Germany). For this purpose, cubes (3 mm × 3 mm) were produced and optically analyzed under temperature influence (25–1000 °C, heating rate 80 °C/min; 1000–1500 °C, heating rate 20 °C/min). The melting point of the sample was assumed to be the flow point, according to DIN 51730 [48]. This refers to the temperature at which the sample has flowed apart to a third of the height it had at a hemispherical temperature.

3.2.4. Bulk Chemistry

The elemental content of the slag samples and the precursors were determined using inductively coupled plasma optical emission spectrometry (ICP-OES 5100, Agilent, Agilent Technologies Germany GmbH and Co. KG, Waldbronn, Germany), e.g., to determine a possible elemental loss (especially Li) during the melting experiment. The specimens were prepared by first preparing a fusion bead with sodium tetraborate in a platinum crucible at a maximum temperature of 1050 °C and then leaching with diluted hydrochloric acid. The results (double determination) were evaluated with a matrix-matched external calibration.

3.2.5. Speciation Analysis and Mineralogical Investigation

The Mn speciation was performed with laboratory-based X-ray absorption near-edge structure analysis (XANES, easyXES100; easyXAFS LLC, Renton, WA, USA). The laboratory instrument features an X-ray tube (W/Pd anode, 100 W, 25 kV, 3 mA), a SDD detector (AXAS-M1, KETEK, Munich, Germany), and a spherically curved crystal analyzer (SBCA, Si 110). For the measurement, pellets (diameter 16 mm) were prepared of 10 mg sample material and 80 mg microcrystalline cellulose (mixed and pressed for 2 min at 10 t). A detailed description of the laboratory-based instrument is given by Wittkowski et al. [11] and Seidler et al. [49]. The energy calibration of the device was performed using a Mn foil (Exafs Materials, Danville, CA, USA). The data were evaluated using a Python script similar to Athena [50]. The Mn edge of precursors, synthetic slag samples and references was scanned from 6482 eV to 6739 eV (step size of 0.25 eV, measuring time of 2 s per step) and from 6739 eV to 6899 eV (step size of 1 eV, measuring time of 1 s per step).

A mineralogical overview of the slag samples was obtained by PXRD (Panalytical X-Pert Pro Diffractometer) using a Co-X-Ray tube (Malvern Panalytical GmbH, Kassel, Germany, λ = 1.7902 Å, 40 kV, 40 mA). A step size of 0.0066° and a time per step of 150.45 s were used (2θ angular range was 5° to 100°). For the assignment of the individual phases, the pdf-2 ICCD XRD database [51] and the American Mineralogist Crystal Structure Database [52] were applied. The evaluated data were compared with the database of materials projects [53].

For element determination within the different phases/single crystals and grains, EPMA (Cameca SX^FIVE FE Field Emission, CAMECA SAS, Gennevilliers Cedex, France) was used. For the quantification of the individual measurement points, the following lines were selected: Na (Kα), Mg (Kα), Al (Kα), Si (Kα), Ca (Kα) and Mn (Kα). Appropriate certified reference materials (CRM: PandH Developments Ltd.; Glossop, Derbyshire, UK and Astimex Standards Ltd.; Toronto, ON, Canada) were used for calibration. For the measurement, the prepared thin sections are coated with a fine carbon layer to ensure the conductivity of the sample. Measurements were performed at 15 kV and a beam size set to 0, with a mean diameter of less than 1 µm (100–600 nm; Schottky-type [54]). The measured intensities of X-rays emitted by the samples were evaluated using the X-PHI model [55].

Light elements such as Li are almost not analyzable by X-ray methods due to the long wavelength of Li Kα as well as the low fluorescence yield. Assuming that there are no other non-detectable light elements in the slag, the content of Li in the respective phases can be determined using virtual components [29]. For an accurate determination of the Li content, it is necessary that the other element concentrations in the grain be measured with the highest possible precision. Repeated measurements of the international standard rhodonite (MnSiO₃; Astimex) were performed with each measurement campaign (

Table 3. The measurements on the CRM rhodonite prove that it is possible to detect the Mn content with ±0.05% accuracy (N = 20 from different days). Indication in wt.%. * The iron content was not measured.

wt.%	Average Rhodonite	%StDev. Rhodonite	Ref. Rhodonite
MgO	1.91	0.02	2.0
Al2O3	0.013	0.004	n.a.
SiO2	46.66	0.08	46.8
CaO	4.78	0.04	4.6
MnO	42.30	0.05	42.3
FeO *	4.34	n.a.	4.3

).

4. Results

A potential element loss during the melt experiment was identified via ICP-OES. An overview of the mineralogical phases in the synthetic slag samples was provided by PXRD. The exact phase composition as well as microstructures were analyzed with EPMA. For an accurate phase assignment, the determination of the Mn oxidation state via XANES measurements was indispensable. Thus, during the following phase assignment via EPMA, Li-manganates with the same Mn content but higher/lower oxidation states were excluded (

Table 4. A selection of different Li-manganates and Mn oxides. The shown compounds have approximately the same Mn concentrations in one column but different oxidation states. For a correct phase assignment, it is important to determine the Mn speciation in the sample. Indication of Mn_total in wt.%.

Phase	Mntotal	Mnox.	Phase	Mntotal	Mnox.	Phase	Mntotal	Mnox.
Li2Mn3O4	67.9	2.0	Li2Mn2O3	64.0	2	LiMnO2	58.5	3.0
LiMn2O3	66.6	2.5	LiMn6O12	62.4	3.5	Li1.1Mn2O4.35	58.7	3.5
LiMn5O8	67.1	3.0	MnO2	63.2	4	Li2Mn4O9	58.2	4.0

).

4.1. Determination of Melting Points

The beginning of vitrification is the temperature at which agglutination occurs at the interface of the particles. With the beginning of deformation, the first signs of softening, such as a change in the surface, occur on the cube. The half-cone point is the point at which the sample is only half as high as its baseline. At the melting point, the temperature is reached at which the sample flows apart. S1 has the highest melting point at 1205 ± 5 °C, S2 has the lowest at 1102 ± 5 °C and the melting point of S3 is 1144.7 ± 5 °C. (Figure 3).

4.2. Bulk Chemistry

The results from the ICP-OES can be taken from

Table 5. Determined mole fraction, given in at.%, of precursor and crystallized slag samples with ICP-OES. The results should clarify whether a loss of elements is to be expected during the melting experiment.

	Raw Mix (at.%)			Product (at.%)			Recovery %
	S1	S2	S3	S1	S2	S3	S1	S2	S3
Li2O	14.3	14.6	13.7	14.0	14.5	13.8	−2.3	−0.6	0.7
MgO	n.a.	2.1	2.3	n.a.	2.2	2.2	n.a.	4.6	−4.3
Al2O3	n.a.	n.a.	1.8	n.a.	n.a.	1.8	n.a.	n.a.	1.3
SiO2	37.4	35.6	35.6	37.7	35.6	35.6	1.0	0.1	0.0
CaO	41.0	40.5	39.3	40.8	40.4	39.3	−0.3	−0.2	−0.1
MnO	7.3	7.3	7.1	7.4	7.4	7.1	1.1	0.2	0.2

. In the three melt experiments the loss of Li was 2.4% (S1), 2.3% (S2) and 3.4% (S3). Compared to the analyzed synthetic slags of Schirmer et al. [13] and Wittkowski et al. [11], a lower Li loss can be documented.

4.3. Speciation Analysis and Mineralogical Investigation

4.3.1. XANES

The reference materials show a shift of the edge to higher energies with an increase in the oxidation state of Mn. The Li-manganates show a characteristic main-edge shoulder. The edge shift and the shape of the curves of the precursors S1 and S3 indicate an oxidation state of Mn²⁺ with a small trend towards Mn^2.67+. Precursor S2 shows greater tendencies towards an oxidation state of Mn^2.67+ (Figure 4).

The K-edges of slag samples S2 and S1 are shifted further to higher energies. Accordingly, the highest average Mn oxidation state should be present in S1. In the spectrum of S1, the characteristic main-edge shoulder of Li-manganate(III) (LiMnO₂) can be detected. Compared to the references, the spectrum of S1 is similar to that of LiMnO₂, with tendencies towards Mn₃O₄. The shifting and shape of the curves S2 and S3 are similar to Mn₃O₄. In relation to the slag samples, the spectrum S3 shows the lowest oxidation state (Figure 5).

4.3.2. PXRD

The results shown in Figure 6 provide an overview of the diffractograms of the three slag samples. All three samples show diffraction peaks that correspond to the phases wollastonite (CaSiO₃), larnite (Ca₂SiO₄) and Li-manganate(III) (LiMnO₂).

The Li-manganate reflexes are best described by the diffraction pattern of LiMnO₂ (ICDD PDF2 00-035-0749) (Figure 6B). Due to line overlays with wollastonite, the presence of hausmannite (ICDD PDF2 00-024-0734) was not clearly verifiable. In the slag-forming phases (wollastonite, larnite), Ca can be replaced by Mn due to comparable ion radii [35]. Therefore, a solid solution between Ca-silicates and Mn-silicates can be expected (Figure 6C).

4.3.3. EPMA

The analysis of single mineral grains/crystals was carried out with EPMA. By means of point measurements, linescans and element distribution patterns, the element contents of individual compounds or grains were analyzed. In addition to the spatial chemical analysis, BSE(Z), micrographs were used to characterize the morphology of the slag samples and to obtain information about the processes during solidification.

The local compositions of the grains visible in the BSE contrast determined by EPMA indicate the following phases:

• Wollastonite: CaSiO₃
• Larnite: Ca₂SiO₄
• Lithium-manganate: Li_(1–x)Mn_(1+0.33x)O₂/Li_(1+x)Mn_(1–0.33x)O₂
• Hausmannite: Mn²⁺Mn³⁺₂O₄
• Amorphous residual melt

Figure 7 shows an overview of the phases present in the slags. The regions of different brightness in the BSE images indicate grains/crystals of different phases with different chemical compositions. The Li-manganates (LiMnO₂) are intergrown with hausmannite (Mn²⁺Mn³⁺₂O₄) in a matrix consisting of Ca and Si, as well as the amorphous residual melt.

Lithium-Manganate (LiMnO₂)

The samples contain an oxide with about 58.5 wt.% Mn and no other measurable elements. After evaluation of all possible (Li)-Mn-oxides under consideration of the speciation analysis with XANES, it should be LiMnO₂. This LiMnO₂ forms 10 to 100 µm large idiomorphic to hypidiomorphic crystals in a Ca-silicate matrix and crystallizes first together with hausmannite. The LiMnO₂ crystals have an average size of 25 µm and are often intergrown with hausmannite (Figure 7). Point measurements inside this phase show the incorporation of traces of Al (S3) (maximum 0.35 wt.%). The elements Ca, Si and Mg were not incorporated. The results of linescans within the crystals indicate that the Mn concentration does not decrease towards the crystal rims (Figure 8A). The path of the linescan is shown in Figure 8B.

The average value of Mn content in the investigated LiMnO₂ grains over all three synthetic slag samples is in S1 58.82 wt.% ± 0.18, in S2 58.67 wt.% ± 0.20 and in S3 58.53 wt.% ± 0.35. In LiMnO₂, the theoretical Mn content is 58.52 wt.% (Figure 9).

Based on the measurement results, for all samples, an average (N = 20) structure formula of LiMnO₂ was obtained [29,56]. With respect to this average, there is a tendency toward a relative Mn enrichment compared with the stoichiometric formula LiMnO₂. In

Table 6. Deviation from the average LiMnO₂ composition. An average structure formula is given, as well as the structural formula based on the lowest (min. Li_(1+x)Mn_(1–0.33x)O₂) and highest (max. Li_(1–x)Mn_(1+0.33x)O₂) Mn concentration.

Sample	LiMnO2 (average)	LiMnO2 (min.)	LiMnO2 (max.)
S1 (N = 20)	Li1.00Mn1.00O2	Li1.00Mn1.00O2	Li0.95Mn1.02O2
S2 (N = 20)	Li1.00Mn1.00O2	Li1.02Mn0.99O2	Li0.97Mn1.01O2
S3 (N = 20)	Li1.00Mn1.00O2	Li1.04Mn0.99O2	Li0.96Mn1.01O2

, the minimum and maximum deviations from ideal LiMnO₂ are listed.

Hausmannite (Mn²⁺Mn³⁺₂O₄) and Spinel (XY₂O₄)

Hausmannite was only detected in samples S1 and S3. The idiomorphic to hypidiomorphic hausmannite grains of S1 and S3 have approximately the same size as the Li-manganates. As shown in Figure 7A,C, A II, hausmannite can occur as individual grains in the matrix or as an intergrowth with LiMnO₂. Within the crystals, an almost homogeneous Mn distribution could be observed. The incorporation of Al (average 1.17 wt.%) and Mg (average 0.69 wt.%) in slag sample S3 as well as Li (calculated using virtual compounds) in small amounts indicates possible spinel solid solutions between Li₂Mn₂O₄, MgAl₂O₄, Mn₃O₄ and MnAl₂O₄. With the measured Mg, Al and Mn concentrations, the following mineral formula (Li_(2x),Mg_(1x),Mn_(2+(1–x)))_1+x(Al_(2–z),Mn³⁺_(z))₂O₄ can be calculated by stoichiometry (

Table 7. The hausmannite spinel solid solution in S3 (N = 18) can be described with MnAl₂O₄, MgMn₂O₄, Li₂Mn₂O₄ and Mn₃O₄. The ratio of Mn^2+/3+ and the content of Li were determined by virtual components. The element fractions of the virtual components can be taken from the table. The structure formula of the compound can be established with the stoichiometric factors x and z.

wt.%	Mn2+Al2O4	MgMn3+2O4	Li2Mn3+2O4	Mn2+Mn3+2O4	Total
Li2O	0.0	0.0	0.8	0.0	0.8
MgO	0.0	1.1	0.0	0.0	1.1
Al2O3	2.2	0.0	0.0	0.0	2.2
MnO	1.5	0.0	0.0	26.6	28.1
Mn2O3	0.0	4.5	4.0	59.1	67.5
total [%]	3.8	5.6	4.7	85.7	99.7
X	0.06	X, Z: stoichiometric factors
Z	1.90	(Li(2x),Mg(1x),Mn(2+(1–x)))1+x(Al(2z),Mn3+(z))2O4

).

For the calculation of the structural formula of hausmannite (Mn²⁺Mn³⁺₂O₄), the Mn²⁺/Mn³⁺ ratio was obtained from the Mn_total concentration. Additionally, spinel-solid solutions were considered in S3 (

Table 8. Calculated structural formula of hausmannite (Mn₃O₄). For S3, the structural formula for the spinel solid solutions ((Li_(2x),Mg_(1x),Mn_(2+(1–x)))_1+x(Al_(2–z),Mn³⁺_(z))₂O₄) was considered. In S2, no hausmannite was found.

Sample	Hausmannite (Mn2+Mn3+2O4)
S1 (N = 20)	Mn2+1.00Mn3+2.00O4
S3 (N = 18)	(Li(0.11)Mg(0.06)Mn2+(0.88))1.06(Al(0.10)Mn3+(1.90))2.00O4

) [11,56].

Matrix (CaSiO₃; Ca₂SiO₄)

The main matrix-forming mineral is wollastonite (CaSiO₃), followed by larnite (Ca₂SiO₄). Both phases form xenomorphic crystals. Within these phases, Mn and Mg are incorporated into the crystal lattice in different concentrations.

Compared to wollastonite, more Mn and Mg are incorporated in larnite. The lowest incorporation is seen in slag sample S3. Wollastonite belongs to the pyroxene-like group; therefore, solid solutions between MnSiO₃ and CaSiO₃ are possible (Figure 10A). Solid solutions between Mn₂SiO₄, Mg₂SiO₄ and Ca₂SiO₄ can occur as well because larnite belongs to the olivine group (Figure 10B).

The structural formulas of wollastonite ((Ca,Mn²⁺)SiO₃) and larnite ((Ca,Mn²⁺)₂SiO₄) were calculated using the virtual compounds MgSiO₃, CaSiO₃, MnSiO₃, Mg₂SiO₄, Ca₂SiO₄ and Mn₂SiO₄ (

Table 9. Overview of the calculated structural formulas for wollastonite (CaSiO₃) and larnite (Ca₂SiO₄).

Sample	Wollastonite CaSiO3	Larnite Ca2SiO4
S1 (N = 22)	(Ca(1.00)Mn2+(0.02))1.02Si0.99O3	Ca(1.73)Mn2+(0.15))1.88Si1.00O4
S2 (N = 18)	(Ca(0.98)Mn2+(0.02))1.00Si1.00O3	(Ca(1.73)Mn2+(0.04))1.77Si1.01O4
S3 (N = 18)	(Ca(0.99)Mn2+(0.02))1.01Si1.00O3	(Mg(0.08)Ca(1.73)Mn2+(0.12))1.92Si1.00O4

) [56].

Amorphous Phase

In every slag sample, there was a residual melt, which was located in interstices and fissures. The elements in this phase were not homogeneously distributed. The concentration of elements (e.g., Mg, Mn) in different grains in the respective slag samples also varies (Figure 11A,B). Thus, it was not possible to calculate a stoichiometric mineral formula for this phase. Furthermore, it was not possible to make statements about the possible Li content within this structure as well as the ratio between Mn^2+/3+. The BSE(Z) image shows where the line scan measurements were taken in the S3 thin section (Figure 11C).

5. Discussion

The aim of this study was to produce a Li and Mn-rich slag and to investigate if it is possible to stabilize a Li-manganate in the melt and if this phase is suitable as a potential new EnAM. In addition, the influence of Mg and Al on the formation of spinel-like Mn oxides was also investigated. The experiments provide an overview of the chemical reactions/phase transformations that take place during the solidification of a slag system based on Ca-silicates. To investigate the influence of Al and Mg impurities, 2 wt.% of Mg was added to S2, and 2 wt.% of Mg and Al was added to S3. However, the results presented should serve as a first impression of which phases crystallize in a complex (Li, (Mg), (Al), Si, Ca, Mn) oxidic slag. The formation, stabilization and purity of possible Li-Mn-O EnAMs in slags, as well as the probable influence of Mg and Al on slag-forming phases, will be discussed in the following section.

5.1. Potential Li-Mn-O EnAM

The crystallized LiMnO₂ forms idiomorphic to hypidiomorphic crystals. The crystal shape indicates that LiMnO₂ (S1–S3), along with hausmannite (S1, S3), is the first to crystallize in the slag samples. Hausmannite and LiMnO₂ are often found intergrown together in one idiomorphic crystal, indicating a decomposition process of an early crystallizate (“pseudomorphosis”). A decomposition of this kind is described by Longo et al. [32] and Paulsen and Dahn [33].

Possible reactions are:

3 LiMn₂O₄ → 3 LiMnO₂ + Mn₃O₄ + O₂

(1)

2 LiMn₂O₄ → LiMnO₂ + LiMn₃O₄ + O₂

(2)

16 Li₂MnO₃ → 4 LiMnO₂ + 4 Mn₃O₄ + 16 Li₂O + 5 O₂

(3)

6 Li₂MnO₃ → 3 LiMnO₂ + LiMn₃O₄ + 4 Li₂O + 2 O₂

(4)

According to this, the Mn in the precursor, which has an average oxidation state of Mn^2.67+ (XANES, Figure 4), must be oxidized to Mn^3.5+ (Equations (1) and (2)) or Mn⁴⁺ (Equations (3) and (4)), respectively, during the high temperature part of the melting experiment. From the observations, these phases decompose during solidification (cooling) with the release of oxygen due to the normal atmospheric conditions (see also Longo et al. [32]). Due to the more or less cubic crystal form of the LiMnO₂ and Mn₃O₄ intergrowths (Figure 7A II), it is reasonable to suggest that a spinel (Equations (1) and (2)) has decomposed. There is an indication that Li and Mn can be replaced to a limited extent by each other within the LiMnO₂, leading to a formula like Li_1–xMn_1+0.33xO₂ or Li_1+xMn_1–0.33xO₂ (Figure 9).

The LiMnO₂ examined has proven to be a pure phase. Only traces of Al (maximum 0.35 wt.%) are incorporated into the crystal lattice in S3. There also exists no solid solution with SiO₂, such as for LiAlO₂ (several wt.%, as observed by [10,11,13]).

5.2. Influence of Mg and Al

According to this study, the addition of Mg has a positive influence on the suppression of hausmannite, as this phase was not found in S2. The addition of Al to S3 causes the formation of hausmannite and, thus, the occurrence of spinel solid solutions. Due to the presence of Mg, Al and Mn at oxidation states Mn²⁺ and Mn³⁺, solid solutions between the spinels Mn₃O₄, MgAl₂O₄ and MnAl₂O₄ can be observed. These solid solutions are caused by the Jahn-Teller effect, which has an influence on the Mn³⁺ above 1172 °C [15]. At temperatures above 1172 °C, the tetragonal crystal system changes to cubic (spinel). The octahedral site in the spinel is occupied by Mn³⁺ and Al³⁺ and the tetrahedral site by Mn²⁺ and Mg²⁺. Generally, the solid solution includes Li₂Mn₂O₄ and could therefore be described by the following formula: (Li_(2x),Mg_(1x),Mn_(2+(1–x)))_1+x(Al_(2–z),Mn³⁺_(z))₂O₄. The Li content was calculated using the virtual compound LiMnO₂. Point measurements and linescans in these small crystals show constant, homogeneous element concentrations. Therefore, a complete solid solution above 1172 °C is assumed. There were no dissolution lamellae, as observed by Wittkowski et al. [11].

The added Al has no influence on the matrix-forming phases (CaSiO₃ and Ca₂SiO₄). In addition to the divalent Mn, Mg can also be incorporated into the crystal lattice of larnite and wollastonite and be located in the lattice position of Ca. Larnite belongs to the olivine group and additionally incorporates Mg and Mn²⁺, which means that there exists a solid solution between the end members Mg₂SiO₄, Ca₂SiO₄ and Mn₂SiO₄. In contrast, wollastonite belongs to the pyroxene-like mineral group. Within this group, solid solutions between MgSiO₃, CaSiO₃ and MnSiO₃ are possible, but only those between CaSiO₃ and MnSiO₃ were identified. The cation position in the larnite is not completely occupied. This could be caused by rapid cooling. The crystal form and elemental composition of two phases, CaSiO₃ and Ca₂SiO₄, suggest crystallization towards the end of solidification.

6. Conclusions and Outlook

The experiments and investigations have shown that it is possible to generate Li-manganate in a complex system (Li₂O-MgO-Al₂O₃-SiO₂-CaO-MnO). It was proven that LiMnO₂ forms as an early crystallizate next to hausmannite (except S2). The manganese speciations are between 2 and 3, which is why part of the Mn²⁺ was incorporated into the matrix. Small amounts of Mg (2 wt.%) have no influence on phase formation. On the other hand, already 2 wt.% Al is sufficient to form spinel solid solutions between Mn²⁺Mn³⁺₂O₄, MnAl₂O₄ and MgAl₂O₄.

The observed Li-manganates only contain traces of other elements (S3: Al ≤ 0.35 wt.%), as is the case with LiAlO₂, which forms a limited solid solution with SiO₂ (2–3 wt.%). In the optimal case, this LiMnO₂ could be directly reused (possibly after recrystallization or pyrolysis). In order to prevent spinel solid solutions and thus to avoid competitive reactions with Li-manganates as potential new EnAM, the aim for further experiments would be either to remove Al from the system or to stabilize higher Mn speciations in the slag, because it is to be expected that, especially in the case of Mn⁴⁺, no reaction with Al occurs.

Continuing studies are investigating whether higher Mn speciations (Mn⁴⁺) in the slag can be stabilized by pure oxygen so that further potential Li-manganate EnAMs (e.g., Li₂MnO₃, LiMn₂O₄) are formed and spinel formation with Al is suppressed. In this case, it is necessary to investigate whether a Li-manganate is formed as an early crystallizate or a Li-aluminate such as LiAlO₂. Furthermore, the influence of Mg and Al (in different concentrations) must be investigated again.

In addition stability fields of phases are to be investigated in the future by means of thermal analysis (e.g., TG-DTA) in order to further optimize the cooling curves. Despite Li not being detectable inside the phases with EPMA, the results of PXRD and XANES show the formation of Li-manganates. For further experiments, spatially resolved mapping of Li with secondary ion mass spectrometry or laser ablation inductively coupled plasma mass spectrometry on the synthetic will be performed.