Structural Understanding of MnO–SiO₂–Al₂O₃–Ce₂O₃ Slag via Raman, ²⁷Al NMR and X-ray Photoelectron Spectroscopies

Abstract

In order to understand sulfur solubility in the MnO–SiO₂–Al₂O₃–Ce₂O₃ slag or inclusion system (mol MnO/mol SiO₂ = M/S = 2.2 or 0.9, Al₂O₃ = 14.2[± 1.8] mol%, Ce₂O₃ = 0–5.6 mol%), the effect of Ce³⁺ ions on the structure of the Mn–aluminosilicate system with different M/S molar ratios has been studied by micro-Raman spectroscopy, solid-state nuclear magnetic resonance (NMR) spectroscopy and X-ray photoelectron spectroscopy (XPS). Because constant oxygen partial pressure was maintained at p(O₂) = 2.8 × 10⁻⁷ atm, oxidation state of Mn and Ce would be in Mn²⁺ and Ce³⁺ in the MnO–SiO₂–Al₂O₃–Ce₂O₃ system, which were also confirmed by XPS analysis. Although it was difficult to measure the ²⁷Al solid-state NMR spectra of the quenched MnO–SiO₂–Al₂O₃ system due to the paramagnetic effect, ²⁷Al NMR spectra did show structural changes in aluminate when cerium was added to the Mn–aluminosilicate melts. Addition of Ce₂O₃ in the high M/S(= 2.2) system caused both an explosive enhancement in the intensity of Raman bands at 600 cm⁻¹ and a peak shift in ²⁷Al NMR spectra, indicating the transition from the [AlO₄]^–:0.5Mn²⁺ tetrahedron to a [AlO₆]^3–:Ce³⁺ octahedron complex due to a strong attraction between aluminate and the cerium ion. XPS spectra show that the transition of the aluminate structure upon introduction of Ce³⁺ ions consumes free oxygen in Mn–aluminosilicate melts in high M/S(= 2.2) system. However, the same structural changes were not observed when Ce₂O₃ was added to lower M/S(= 0.9) system, because Ce³⁺ ions primarily interact with the pre-existing [AlO₆]-octahedron in low M/S system.

Graphic Abstract

1 Introduction

It is very important to control sulfur contents in molten steel because the uptake of sulfide or oxy-sulfide type inclusions deteriorates the corrosion resistance and mechanical behavior of steels [1, 2]. Sulfur could be effectively removed from molten steel by a liquid oxide slag and/or inclusion system containing rare earth elements, which has a high solubility for sulfur because of strong attraction between rare earth elements and sulfur at temperatures that are typical for steelmaking.

Anacleto et al. [3] investigated the effect of Ce oxide on the sulfide capacity of the CaO–SiO₂ slag from the slag-metal reaction between CaO–SiO₂–Ce₂O₃ slag and carbon saturated iron at 1773 K. It was concluded that the addition of Ce oxide to CaO–SiO₂ slag is more effective on desulfurization than that of CaO because it decreases the activity coefficient of SiO₂ in the slags. Yang et al. [4] presented that substituting Ce₂O₃ for Al₂O₃ in the CaO–Al₂O₃–SiO₂–MgO (CaO/Al₂O₃ = 1.5, SiO₂ = 10 wt%, MgO = 7 wt%) slag has positive effect on desulfurization in Al-killed steel at 1873 K because increase in Ce₂O₃ content improves the fluidity of molten slags. Nevertheless, it is not clear to understand the effect of Ce₂O₃ on sulfide capacity of the slags in a thermodynamics standpoint in these previous studies.

In our previous study [5], the sulfide capacity of the MnO–SiO₂–Al₂O₃–Ce₂O₃ (MnO/SiO₂ = 0.3–2.2 molar ratio) quaternary system was measured by the gas-slag equilibration method at 1873 K in order to evaluate the effect of Ce oxide on the sulfur speciation behavior in the quaternary system. The addition of Ce₂O₃ to the MnO–SiO₂–Al₂O₃–Ce₂O₃ quaternary system at high basicity compositions (i.e., MnO/SiO₂ > 1, molar ratio) decreased the sulfide capacity of the quaternary system unlike the results of Anacleto et al. and Yang et al. Because there is limited thermodynamic information for the cerium oxide containing slags such as activity of components, phase diagrams, etc., the short range order structural units of the system was analyzed using micro-Raman spectroscopy, which was widely employed to investigate MnO-containing silicate melts [6,7,8,9], in order to understand the role of Ce₂O₃ for the solubility of sulfur [5]. Nevertheless, the structure of the MnO–SiO₂–Al₂O₃–Ce₂O₃ system poses more open questions that warrant further investigation.

Many researchers sought to investigate oxide melts, glasses and crystals containing transition metal or rare earth cations using solid-state nuclear magnetic resonance (NMR) spectroscopy [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25]. The presence of paramagnetic cations significantly broadens NMR peaks, accompanied by a loss of peak intensity [14, 15, 21]. Kim et al. [15, 21] propose that the spectra of iron-bearing glass do not provide clear resolution of multiple Si environments because the intensity of peaks from iron-bearing glass decrease and broaden with increasing Fe₂O₃ content. Signal loss and peak broadening are primarily due to a strong effect of the unpaired electrons in the d-orbitals of Fe on the decrease of spin–lattice relaxation time of Si nuclear [15, 21].

The strong interaction between the unpaired electrons in the paramagnetic cation and the observed nuclear spins could either positively and negatively shift the NMR peaks [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25]. It has been discussed that the paramagnetic shifts are caused by Fermi contact shift (through-bond transfer of unpaired electron spin density) and/or pseudo contact shift (through-space dipolar coupling from a magnetic cation in an asymmetrical site) [10,11,12, 16,17,18, 22,23,24]. It is observed that the number of unpaired electrons for M cation (M = Mn²⁺, Fe²⁺, Co²⁺, Ni²⁺) strongly affects the increase in paramagnetic shifts in ³¹P or ²⁹Si NMR peaks for Li-phosphate olivine (LiMPO₄) or silicate olivine (M₂SiO₄) [13, 19, 25]. For investigating the structure of Ce doped yttrium aluminum garnet (YAG) using ²⁷Al NMR spectroscopy, Ce³⁺ cations, which are neighbored near [AlO₄] and [AlO₆]-units, also cause the paramagnetic shifts in the ²⁷Al NMR peaks [18, 22, 24]. Therefore, it is expected that the manganese and cerium cations in the MnO–SiO₂–Al₂O₃–Ce₂O₃ system would result in the paramagnetic shifts in ²⁷Al NMR peaks.

Because the radius of Ce³⁺ ion (r = 1.01 Å) is too large to be a network former such as [SiO₄]- tetrahedrally coordinated units and is similar to that of Ca²⁺ (r = 1.0 Å) or Na⁺ (r = 1.02 Å), Lin et al. [26, 27] proposed that the Ce³⁺ ion has a preference to be charge compensator or network modifier. Wu and Pelton [28] reported strong interaction between Ce₂O₃ and Al₂O₃ from the evaluation of phase diagram for R₂O₃–Al₂O₃ (R: Rare earth) binary systems. It was also revealed that the activity coefficient of Al₂O₃ declines with the increase in Ce₂O₃ content in the CaO–Al₂O₃ melts at 1773 K due to a strong attraction between Ce₂O₃ and Al₂O₃ [29, 30]. Hence, it is conclusive that the Ce₂O₃ has a strong interaction to Al₂O₃ in the present Mn–aluminosilicate system.

In the present study, structural analysis of the MnO–SiO₂–Al₂O₃–Ce₂O₃ quaternary system with different MnO/SiO₂ (M/S = 0.9 and 2.2) ratio has been systematically carried out using micro-Raman spectroscopy, ²⁷Al magic angle spinning-nuclear magnetic resonance (MAS-NMR) spectroscopy and X-ray photoelectron spectroscopy (XPS) to elucidate the effect of cerium oxide on the structural changes in Mn–aluminosilicate system quenched from 1873 K.

2 Experimental Procedures

2.1 Preparation of Oxide Samples

A super-kanthal vertical electric furnace was applied to prepare the MnO–SiO₂–Al₂O₃–Ce₂O₃ melt at 1873 K. A proportional integral differential controller connected with B-type thermocouple was used in the vertical furnace to control the temperature. Before all of the experiments was carried out, the temperature inside the furnace was checked using an external B-type thermocouple. The mixtures of reagent-grade MnO, SiO₂, Al₂O₃, and CeO₂ were used to manufacture the glass samples. The oxide mixture (1.2 g) in a platinum crucible held in a porous alumina holder was positioned at the hot zone of the furnace under CO–CO₂–Ar gas mixture (total flow rate = 400 mL/min) for 12 h. Constant oxygen partial pressure was maintained at p(O₂) = 2.8 × 10⁻⁷ atm using a CO–CO₂ reaction, i.e., CO(g) + 1/2O₂(g) = CO₂(g), during the equilibration of the oxide melt with the gas mixture at 1873 K.

After 12 h, the platinum crucible containing the oxide mixture was quenched by dipping it into brine. The quenched samples were ground under 100 μm using a crusher for spectroscopic analysis. The compositions of the quenched samples were confirmed using a Bruker S4 explorer X-ray fluorescence spectrometer and the results are listed in Table 1. X-ray diffraction (XRD, D/Max-2500/PC, Rigaku) analysis was carried out to confirm whether the samples were glassy (amorphous) or not. Monochromatized Cu k-alpha radiation was used as the X-ray sauce (λ = 1.5418 Å) and operated at 40 kV and 100 mA. One representative XRD result is shown in Fig. 1, from reflecting the amorphous nature of all the samples were homogeneous liquids at 1873 K.

Table 1 Experimental composition of oxide samples (mol%)

Fig. 1

Typical XRD pattern of the quenched MnO–SiO₂–Al₂O₃–Ce₂O₃ (MnO/SiO₂ = 2.2) glass sample

2.2 Structural Analysis of Quenched Samples Using Spectroscopic Methodologies

Micro-Raman spectroscopic analysis was carried out using a ‘LabRam Aramis’ Horiba Jobin–Yvon spectrometer. The 514-nm Ar-ion laser source was used for Raman scattering measurements of the quenched oxide samples. Further details on the preparation, measurement procedure and data processing are available in our previous articles [6,7,8,9]. The ²⁷Al NMR spectra were collected using a Bruker Avance II 500 MHz solid-state NMR spectrometer at 11.7 T with a 4-mm-thick wall zirconia rotor and a Doty Scientific double-resonance MAS probe at a spin rate of 10 kHz. The ²⁷Al NMR spectra were collected at radio-frequency fields of 130.32 MHz and with single pulse acquisition with a pulse length of 2.0 μs (radio frequency tip angle of 30°). The recycle delay was 1 s for all samples.

The Mn2p, Ce3d and O1s XPS spectra of the quenched oxide samples were measured using an XPS analyzer (K-alpha by Thermo UK). Monochromatized Al K-α radiation (1486.6 eV) was used as the X-ray source and operated at 3 mA and 12 kV. The rod-shaped samples were fractured in ultrahigh vacuum (4.8 × 10⁻⁷ Pa). The surface charge was neutralized by flooding electrons. In order to correct for the charging effects, measured binding energies were calibrated using the binding energy of the adventitious C1s line assuming that the C1s core level is 284.6 eV.

It was confirmed that the oxidation state of Mn would be Mn²⁺ in the present MnO–SiO₂–Al₂O₃ system from the observation of Mn²⁺ peaks (642 eV and 654 eV) and MnO satellite feature in Mn2p XPS spectra in Fig. 2a. It was also confirmed that the cerium ion is stabilized as Ce³⁺ in the present system under the reducing atmosphere at 1873 K because the intensity of Ce³⁺ peaks (886.4 and 904.3 eV) is majorly detected in conjunction with relatively small intensity of Ce⁴⁺ peaks (916 eV) in Fig. 2b.

Fig. 2

The XPS spectra of a Mn 2p and b Ce 3d core level for quenched oxide sample

A typical deconvolution result of the O1s binding energy (eV) from XPS spectra for the MnO–SiO₂–15.4Al₂O₃–3.7Ce₂O₃ (mol%, MnO/SiO₂ = M/S = 2.2) system is shown in Fig. 3. On the basis of the Gaussian function, the O1s XPS spectra have been deconvoluted into three peaks at about 530 eV, 531 eV and 532 eV, which correspond to free oxygen (O²⁻), non-bridging oxygen (O⁻) and bridging oxygen (O⁰), respectively [31,32,33]. The free oxygen is defined as binding energy of MnO in the O1s XPS spectra because O²⁻ ions has ionic bonding with MnO in the MnO–SiO₂–Al₂O₃–Ce₂O₃ system. The commercial software ‘PeakFit’ was used for deconvolution of the spectra with an r² value over 0.99.

Fig. 3

A typical Gaussian deconvolution result of the O1s binding energy (eV) for the quenched MnO–SiO₂–Al₂O₃–Ce₂O₃ (MnO/SiO₂ = 2.2) system

3 Results and Discussion

3.1 Effect of MnO on ²⁷Al NMR Spectroscopy in MnO–SiO₂–Al₂O₃ System

In order to understand the effect of the paramagnetic cation, i.e. Mn²⁺, on ²⁷Al NMR peaks, ²⁷Al NMR spectra of the MnO–SiO₂–Al₂O₃ system (MnO/SiO₂ = 0.9 molar ratio, Al₂O₃ = 14 mol%) are shown in Fig. 4 in conjunction with ²⁷Al NMR spectra of the CaO–SiO₂–Al₂O₃ system (CaO/SiO₂ = 1 molar ratio, Al₂O₃ = 13 mol%). It is difficult to distinguish the ²⁷Al NMR spectra in the MnO–SiO₂–Al₂O₃ system in Fig. 4a because the unpaired electrons in manganese cation decreases spin–lattice relaxation times in Al nuclear, resulting in peak broadening in conjunction with intensity loss.

Fig. 4

a ²⁷Al NMR spectra and b normalized ²⁷Al NMR spectra of both CaO–SiO₂–Al₂O₃ and MnO–SiO₂–Al₂O₃ systems

McMillan et al. [34] investigated the environment of Al in the CaO–Al₂O₃ binary system of different alumina content using ²⁷Al NMR spectra. They presented isotropic chemical shifts at about 75 ppm, 44 ppm and 13 ppm that correspond to [AlO₄], [AlO₅] and [AlO₆] structural units, respectively. Neuville et al. [35] investigated the structure of aluminate in the CaO–SiO₂–Al₂O₃ system via application of ²⁷Al high-resolution solid-state NMR spectroscopy in conjunction with Raman spectroscopy. They propose that the isotropic chemical shifts in the range 67–71 ppm, 34–40 ppm and 7–11 ppm correspond to [AlO₄], [AlO₅] and [AlO₆] structural units, respectively, and that the proportion of [AlO₄] is at least 87.4% in the CaO–SiO₂–Al₂O₃ system (CaO/SiO₂ = 1, Al₂O₃ = 3–33 mol%). ²⁷Al NMR spectra of the CaO–SiO₂–Al₂O₃ system in Fig. 4a shows a maximum peak at about 60 ppm with a shoulder at about 30 ppm suggesting that the coordination number of Al atoms is 4 by majority with a small portion of [AlO₅] units. This is similar to the results of Neuville et al. [35] in that the [AlO₄] species are coexistent with few [AlO₅] species in the CaO–SiO₂–Al₂O₃ system (CaO/SiO₂ = 1, Al₂O₃ = 14 mol%).

The normalized ²⁷Al NMR spectra of both CaO–SiO₂–Al₂O₃ and MnO–SiO₂–Al₂O₃ systems are shown in Fig. 4b. The chemical shifts of the CaO–SiO₂–Al₂O₃ have maximum peaks at around 60 ppm in ²⁷Al NMR spectra, while the chemical shifts of MnO–SiO₂–Al₂O₃ peak at around − 25 ppm in ²⁷Al NMR spectra. For manganese containing Li-phosphate olivine crystals, the manganese cation strongly causes the paramagnetic shifts in ⁷Li and ³¹P NMR spectra [13, 19]. Because it was clearly revealed by the present authors in previous studies that the Mn²⁺ ion in aluminosilicate glass acts as a network modifier as does Ca²⁺ ion [6,7,8,9], the environment of Al atoms in the aluminosilicate networks of the MnO–SiO₂–Al₂O₃ system may be similar to that in the CaO–SiO₂–Al₂O₃ system. Therefore, we can presume that isotropic chemical shifts at around 10 ppm correspond to [AlO₄], while the chemical shifts at around − 25 ppm correspond to highly coordinated aluminum, [AlO₅] or [AlO₆], in Mn–aluminosilicate glass.

3.2 Structural Changes of MnO–SiO₂–Al₂O₃ System with Different MnO/SiO₂ Ratios

The Raman bands of the MnO–SiO₂–Al₂O₃ system with different MnO/SiO₂ (= M/S) ratios are shown as a function of wavenumber in Fig. 5. The Raman bands in the 850–1200 cm⁻¹ range of the M/S = 2.2 (molar ratio) system exhibit a maximum near 900 cm⁻¹ with a shoulder at about 1000 cm⁻¹. The Raman bands at 900–1100 cm⁻¹ increase with increasing SiO₂ content. The Raman bands at 850–1100 cm⁻¹ in aluminosilicate melts and glasses are assigned to asymmetric Si–O stretching vibration; that is, Raman bands at 850–880 cm⁻¹, 900–920 cm⁻¹, 950–1000 cm⁻¹, and 1050–1100 cm⁻¹ correspond to Q⁰_Si (NBO/Si = 4), Q¹_Si (NBO/Si = 3), Q²_Si (NBO/Si = 2), and Q³_Si (NBO/Si = 1), respectively [6,7,8,9, 36,37,38].

Fig. 5

Raman scattering of the MnO–SiO₂–Al₂O₃ system with different MnO/SiO₂ ratio

Park et al. [6,7,8,9] investigated the structure–property (viscosity, capacity, density and electrical conductivity) relationships of the CaO–MnO–SiO₂ system with different SiO₂ content by analyzing the Raman spectra. They proposed that the substitution of Mn²⁺ for Ca²⁺ results in a decrease of the Q³/Q² ratio, which typically presents the degree of polymerization of the silicate networks, especially in the wollastonite primary system. The Raman bands around 950–1100 cm⁻¹ representing Q² and Q³ in the M/S = 0.9 system become wider than those in the M/S = 2.2 system (Fig. 5), indicating that the more polymerized silicate networks, as expected, increase the contributions to the high energy side of the 950–1100 cm⁻¹ envelop in the glass with higher SiO₂ and lower modifier oxide additions.

Furthermore, the Raman band at 500–600 cm⁻¹ in the M/S = 0.9 system is relatively wide, while the Raman bands at about 600 cm⁻¹ in the M/S = 2.2 system exhibit a single narrow peak. Several authors report that these Raman bands are corresponding to the bending motion of bridging oxygen within T–O–T (T = Si or Al) linkages [36,37,38,39,40]. The broadening and rising of Raman bands at 400–600 cm⁻¹ with a decrease in the M/S ratio is caused by an increase in the angle of T–O–T linkages, which means the number of bridging oxygen within T–O–T become larger in M/S = 0.9 than M/S = 2.2.

3.3 Effect of Ce₂O₃ on Structure of Mn–Aluminosilicate: MnO/SiO₂ = 2.2 (Molar Ratio) System

For corundum (α-Al₂O₃) and boehmite (γ-AlOOH) crystals, which are consist of that Al atoms are octahedrally coordinated, the Al–O stretching vibration of [AlO₆]-octahedron is observed in the range of 500–600 cm⁻¹ Raman bands [41,42,43]. Tarte [42] investigated infra-red (IR) spectra of the crystal structure of aluminate compounds and proposed that the IR band at 750–900 cm⁻¹ correlates with the stretching vibration of the edge sharing [AlO₄]-tetrahedral and the IR band at 650–800 cm⁻¹ correlates with the stretching vibration of corned sharing [AlO₄]-tetrahedral unit. In the case of [AlO₆]-octahedron, the IR bans shifts from 500–680 cm⁻¹ range to 400–530 cm⁻¹ range when the edge sharing [AlO₆]-octahedron is changed to the corner sharing [AlO₆]-octahedron. Poe et al. [44, 45] observed the IR absorption bands of CaAl₂O₄ glass with a maximum near 820 cm⁻¹ and a shoulder near 680 cm⁻¹ due to the stretching vibration of the [AlO₄]-tetrahedral unit. The Raman bands at 730, 780 and 850 cm⁻¹ were associated with the Al-O stretching vibration in the [AlO₄]-tetrahedral units with Q²_Al (NBO/Al = 2), Q³_Al (NBO/Al = 1) and Q⁴_Al (NBO/Al = 0), respectively, in a low-SiO₂ Ca-aluminosilicate system [39, 40].

The effect of cerium addition on the Raman spectra of the MnO–SiO₂–Al₂O₃ (M/S = 2.2, Al₂O₃ = 15 mol%) system is shown in Fig. 6. In the MnO–SiO₂–Al₂O₃ ternary system, Raman bands in the 800–1100 cm⁻¹ range exhibit a peak near 900 cm⁻¹ and a shoulder near 1000 cm⁻¹. Because Raman bands with maximum peak near 900 cm⁻¹ indicate that relative fractions of depolymerized silicate units (i.e., Q⁰_Si [850–880 cm⁻¹] and Q¹_Si [900–920 cm⁻¹]) are in majority, it can be stated that the silicate networks are depolymerized in the present ternary system. Moreover, the Raman band near 600–700 cm⁻¹ is assigned to the depolymerized [AlO₄]-tetrahedral unit, which indicates that the aluminate networks are also depolymerized in the Ce-free MnO–SiO₂–Al₂O₃ (M/S = 2.2, Al₂O₃ = 15 mol%) ternary system.

Fig. 6

Effect of Ce₂O₃ addition on the Raman scattering of the Mn–aluminosilicate system at MnO/SiO₂ = 2.2

The shape of the Raman bands at 800–1100 cm⁻¹, assigned to Si–O stretching vibration with different NBOs, does not change irrespective of cerium content, indicating that cerium oxide does not affect the silicate networks in Mn–aluminosilicates. On the other hand, a sharp increase in Raman band intensity at around 600 cm⁻¹ corresponding to the [AlO₆]-octahedral unit indicates that Ce₂O₃ strongly interacts with the aluminate unit in the Mn–aluminosilicate system. This represents a drastic change in the structure of aluminate units irrespective of changes in alumina and silica content with increasing cerium content.

The aluminate anion mainly exists as [AlO₄]-tetrahedron in a high M/S(= 2.2) system, in which it is balanced with a half mole of charge balancing Mn²⁺ cations (i.e., expressed as hypothetical [AlO₄]^–:0.5Mn²⁺ unit similar to [SiO₄]-unit) as schematically shown in Fig. 7a. However, when Ce₂O₃ is added, the [AlO₆]-octahedron becomes more stable than the [AlO₄]-tetrahedral unit because Ce³⁺ ion has a preferential charge balance role to [AlO_n]-units, resulting in formation of the hypothetical [AlO₆]^3–:Ce³⁺ unit as shown in Fig. 7b. The interaction between ionic species is presumably expressed by the following Eq. (1).

$$ \left[ {{\text{AlO}}_{4} } \right]^{-}{:}0.5{\text{Mn}}^{2 + } + {\text{Ce}}^{3 + } + 2{\text{O}}^{{2{-}}} = \left[ {{\text{AlO}}_{6} } \right]^{{3{-}}} {:} {\text{Ce}}^{3 + } + 0.5{\text{Mn}}^{2 + } $$

(1)

Fig. 7

Schematics of the ionic structure of a the [AlO₄]-tetrahedron balanced with 0.5Mn²⁺ ion (hypothetically shown as [AlO₄]⁻:0.5Mn²⁺ complex) and b the [AlO₆]-octahedron balanced with Ce³⁺ ion (hypothetically shown as [AlO₆]³⁻:Ce³⁺ complex)

Tarte [42] presented IR bands at 400–680 cm⁻¹ as corresponding to the stretching vibration of [AlO₆]-octahedron species in the aluminate crystal. Moreover, Okuno et al. [38] recently observed the band at 600–700 cm⁻¹ in aluminosilicate melts and glasses, which is due to the symmetric stretching vibrations of [AlO₆]-octahedron species. Therefore, the significant increase in the relative intensity of Raman bands at about 600 cm⁻¹, as shown in Fig. 6, with increasing cerium content can be attributed to the strong affinity between Al₂O₃ and Ce₂O₃, as shown in Eq. (1).

²⁷Al NMR spectra of the MnO–SiO₂–Al₂O₃–Ce₂O₃ system are shown in Fig. 8 with varying Ce₂O₃ content from 0 to 5.6 mol%. The peak of ²⁷Al NMR spectra moves from around 0 to − 25 ppm as Ce₂O₃ is added to the MnO–SiO₂–Al₂O₃. For Ce-doped YAG crystals, paramagnetically shifted peaks for [AlO₆] were observed at around − 15 and − 30 ppm in ²⁷Al MAS NMR spectra resulting from neighbored Ce³⁺ cation near aluminum nuclei [18, 22, 24]. The introduction of Ce³⁺ cation to the MnO–SiO₂–Al₂O₃ system modifies the [AlO₄]-tetrahedron balanced by Mn²⁺ cation into the [AlO₆]-octahedron balanced by Ce³⁺ cation as shown in Eq. (1). Therefore, an increase in the shifts of ²⁷Al NMR spectra from − 10 to − 25 ppm with addition of Ce₂O₃ indicates the formation of the [AlO₆]^3–:Ce³⁺ octahedron in the present MnO–SiO₂–Al₂O₃–Ce₂O₃ (M/S = 2.2) system.

Fig. 8

Effect of Ce₂O₃ addition on ²⁷Al NMR spectra of the Mn–aluminosilicate system at MnO/SiO₂ = 2.2

The percentage of oxygen species, viz. free oxygen (O²⁻), non-bridging oxygen (O⁻), and bridging oxygen (O⁰), obtained by XPS analysis for the MnO–SiO₂–Al₂O₃–Ce₂O₃ (M/S = 2.2) system is plotted against cerium oxide content in Fig. 9. Non-bridging oxygen represents the majority due to the availability of network-modifying Mn²⁺ cations, and this majority increases with increasing content of cerium oxide, and vice versa for free oxygen. The percentage of bridging oxygen is independent of cerium oxide content. Consequently, in the high M/S(= 2.2) system, the pre-existing [AlO₄]^–:0.5Mn²⁺ units (Fig. 7a) could be converted to [AlO₆]^3–:Ce³⁺ units (Fig. 7b) by employing free oxygen as the content of Ce³⁺ ions increase as given in Eq. (1).

Fig. 9

Effect of Ce₂O₃ addition on the percentage of oxygen species from O1s XPS in the Mn–aluminosilicate system at MnO/SiO₂ = 2.2

3.4 Effect of Ce₂O₃ on Structure of Mn–Aluminosilicate: MnO/SiO₂ = 0.9 (Molar Ratio) System

The effect of cerium addition on the Raman spectra of the high-silica Mn–aluminosilicate system (i.e., M/S = 0.9, Al₂O₃ = 14 mol%) is shown in Fig. 10. When Ce₂O₃ is added to the M/S = 0.9 system, no critical changes are observed in Raman spectra in the 800–1200 cm⁻¹ range corresponding to Si–O stretching vibration because Ce³⁺ cations rarely affect silicate networks, as mentioned previously. Raman spectra at 600–800 cm⁻¹, assigned to Al–O stretching vibration, is not also significantly changed by cerium addition, which is contrary to the experimental results in the low-silica Mn–aluminosilicate system (M/S = 2.2, Al₂O₃ = 15 mol%) as discussed in Sect. 3.3. As discussed in Sect. 3.1, the existence of highly coordinated aluminum other than the [AlO₄]-unit was noted in the M/S = 0.9 system from the analysis of ²⁷Al NMR spectra. Because Ce³⁺ cation balances [AlO₆]-octahedron is more advantage than [AlO₄]-tetrahedron in view of charge balancing, the Ce³⁺ cation preferentially reacts with pre-existing [AlO₆]-units in the high-silica Mn–aluminosilicate system. Therefore, the addition of Ce₂O₃ to M/S = 0.9 system results in minimal changes to Raman bands assigned to [AlO₆]-octahedron units (Fig. 10).

Fig. 10

Effect of Ce₂O₃ addition on the Raman scattering of the Mn–aluminosilicate system at MnO/SiO₂ = 0.9

The ²⁷Al NMR spectra of the MnO–SiO₂–Al₂O₃–Ce₂O₃ system are shown in Fig. 11 by Ce₂O₃ concentration. The ²⁷Al NMR peak for the MnO–SiO₂–Al₂O₃ ternary system without Ce₂O₃ is located near − 25 ppm, which indicates that [AlO₆]-octahedron units are mainly distributed in the high-silica Mn–aluminosilicate system. Because [AlO₆]-octahedron already exists in the Ce-free ternary system, Ce³⁺ does not need to convert [AlO₄]^–:0.5Mn²⁺ tetrahedron units to [AlO₆]^3–:Ce³⁺ octahedron units in the M/S = 0.9 system. Consequently, the structure of the MnO–SiO₂–Al₂O₃–Ce₂O₃ (M/S = 0.9, Al₂O₃ = 14 mol%) system is minimally changed by cerium addition as experimentally confirmed by Raman and ²⁷Al NMR spectra.

Fig. 11

Effect of Ce₂O₃ addition on ²⁷Al NMR spectra of the Mn–aluminosilicate system at MnO/SiO₂ = 0.9

The deconvolution results of the O1s peak in the MnO–SiO₂–Al₂O₃–Ce₂O₃ (M/S = 0.9) system are shown in Fig. 12 as a function of Ce₂O₃ content. The percentage of free oxygen (O²⁻) in this system is significantly lower than that of the M/S = 2.2 system (Fig. 9), whereas the percentage of bridging oxygen (O⁰) is much higher than that of the M/S = 2.2 system (Fig. 9) because the aluminosilicate networks are more polymerized in high-silica Mn–aluminosilicate system. The relative percentages of the three oxygen species present are nearly constant irrespective of Ce₂O₃ content, which indicates that Ce₂O₃ has minimal effects on aluminosilicate networks in the high-silica system.

Fig. 12

Effect of Ce₂O₃ addition on the percentage of oxygen species from O1s XPS in the Mn–aluminosilicate system at MnO/SiO₂ = 0.9

The Al–O bond mainly has an octahedral coordination under the condition of deficiency of charge balancing cations, i.e., in high-silica glass [34, 35, 45,46,47,48]. Before adding Ce₂O₃, the concentration of [AlO₆]-units in the M/S = 0.9 system are relatively greater than in the M/S = 2.2 system. Thus, Ce³⁺ ions added to high-silica Mn–aluminosilicate system instantly interact with large amounts of [AlO₆]-octahedron units to form more stable [AlO₆]^3–:Ce³⁺ units due to strong affinity between Ce³⁺ and Al³⁺ in [AlO₆]-units (Fig. 7b). Therefore, the structure of the high-silica Mn–aluminosilicate system was not significantly disturbed by the addition of Ce₂O₃.

4 Conclusions

The structure of the MnO–SiO₂–Al₂O₃–Ce₂O₃ systems with different MnO/SiO₂ (= M/S) ratios has been analyzed using micro-Raman spectroscopy, ²⁷Al NMR spectroscopy and XPS spectroscopy. In the present study, high-resolution solid-state NMR spectroscopy was used to elucidate the effect of Ce₂O₃ on aluminate configuration in a Mn–aluminosilicate system. The polymerization of silicate networks in the MnO–SiO₂–Al₂O₃–Ce₂O₃ system is strongly dependent on the M/S ratio. Because Ce³⁺ ion has a strong affinity with aluminate in aluminosilicate melts, the [AlO₄]^–:0.5Mn²⁺ units are transformed to [AlO₆]^3–:Ce³⁺ units by employing free oxygen when Ce₂O₃ is added to the low-silica Mn–aluminosilicate system (M/S = 2.2). Thus, significant changes in aluminate structure occur upon the addition of Ce₂O₃ in the M/S = 2.2 system. However, the aluminosilicate networks of high-silica Mn–aluminosilicate systems (M/S = 0.9) are highly polymerized in conjunction with the existence of [AlO₆]-octahedron units due to lack of a charge compensator, i.e., Mn²⁺. The Ce³⁺ ion has strong interactions with pre-existing [AlO₆]-octahedron units when Ce₂O₃ is added to this system. Therefore, structural changes were not observed despite the addition of Ce₂O₃ to Mn–aluminosilicate (M/S = 0.9) system.