Synthesis of Magnesia–Hercynite-Based Refractories from Mill Scale and Secondary Aluminum Dross: Implication for Recycling Metallurgical Wastes

Abstract

This study investigates the synthesis of magnesia–hercynite-based refractories using blends of magnesia powder, aluminum dross (AD), mill scale (MS), and graphite, focusing on the effects of carbon concentration and heating temperature. The results demonstrate successful synthesis at 1550 °C and 1650 °C, with high magnesia content (C80 and D80) leading to the formation of distinct phases, including MgO, FeAl₂O₄, MgFeAlO₄, CaMg(SiO₄), and Ca₃Mg(SiO₄)₂, which influence the ceramic’s microstructure and mechanical properties. Increased magnesia content reduces porosity and enhances crushing strength, while heating to 1650 °C significantly improves densification and nearly doubles cold crushing strength, from 43.77–58.97 MPa at 1550 °C to 76.79–95.67 MPa at 1650 °C. These findings suggest that the synthesized refractories exhibit properties comparable to commercial magnesia–hercynite bricks, with potential for the further development for industrial rotary kiln applications.

1. Introduction

High alumina refractories were widely used in the high-temperature zones of cement rotary kilns. As kiln temperatures and capacities increased, magnesia–chrome bricks became the preferred lining because of their durability and ability to retain protective coatings [1]. However, concerns over the environmental hazards of hexavalent chromium (Cr⁶⁺) led to the decline in magnesia–chrome bricks, which were replaced by chrome-free alternatives [1,2,3]. Chromite was initially used for its flexibility, but its environmental impact in the sintering zone prompted a shift to iron–alumina or hercynite spinel as safer options. Hercynite, though rare, is primarily produced through electro-fusion, a process that, while effective, is both time- and energy-consuming [4,5,6,7].

Magnesia–hercynite bricks have gained significant attention in the refractory materials industry because of their enhanced mechanical properties, thermal shock resistance, and environmental benefits. The incorporation of hercynite spinel (FeAl₂O₄) into magnesia-based refractories improves their thermo-mechanical properties, including density, porosity, and mechanical strength, making them suitable for demanding applications like cement kilns [8]. Studies show that adding spinel structures such as hercynite promotes microcracking from thermal expansion mismatch, enhancing crack propagation resistance [9]. The generated spinel phases, such as MgFe₂O₄, contribute to a more compact microstructure, especially with fused hercynite. Research on periclase–hercynite bricks under various atmospheric conditions reveals that iron ion diffusion in oxidizing atmospheres causes ring cracks, while more severe diffusion in CO atmospheres can lead to structural damage [10]. Recent advancements, including plasma fusion synthesis, have made the production of magnesia–hercynite bricks faster, more economical and higher performing [4]. Hercynite spinel is emerging as an environmentally friendly alternative to magnesia–chrome bricks, reducing the environmental risks associated with Cr⁶⁺ compounds. Overall, these bricks offer superior mechanical properties and thermal stability, with ongoing research expanding their potential applications [4].

Mill scale (MS) is one of the by-products of the iron and steel industries, generated during the hot rolling process of steel slabs, blooms, and billets. It is notable for its high iron content, comprising over 70% metallic iron or more than 90% iron oxides [11,12]. MS has been widely recycled in steelmaking processes, such as being blended with fossil fuels (coal, coke, and graphite) to replace iron ore or scrap iron [13,14]. Research has explored the direct reduction of MS using various reducing agents, including biomass and reducing gases, to produce iron powder and iron-bearing compounds [15,16,17,18,19].

Primary aluminum dross floats on the surface of molten aluminum during the smelting process. It acts as a protective layer that absorbs impurities and minimizes heat loss. It mainly consists of aluminum oxide and some metallic aluminum. Primary aluminum dross is generally re-melted in a rotary kiln to recover the remaining metallic aluminum. This re-melting process produces secondary aluminum dross (AD), primarily composed of aluminum oxide and salts. AD is often utilized in the production of high-value products such as ceramics, refractory materials, construction materials, and calcined alumina [20,21].

Magnesia–hercynite-based refractory is generally produced by mixing magnesia, or periclase (MgO), with hercynite spinel (FeAl₂O₄) [22,23]. The spinel can be either naturally occurring or synthesized through various methods, such as electro-fusion and plasma fusion [5,6,7,8], using raw materials like alumina powder [4,5], metallic aluminum [7], iron oxide [5,8], and mill scale [7].

Because of the limited availability of natural or virgin material resources, reducing the consumption of these materials in various industrial processes is essential. Rather than relying on virgin materials, this study aims to utilize industrial wastes, specifically mill scale and aluminum dross, as alternative sources of iron oxides and alumina for the production of magnesia–hercynite refractory. This research focuses on synthesizing magnesia–hercynite refractory by blending magnesia powder, mill scale, and aluminum dross in a single-step process, using graphite as a reducing agent in a carbothermic reduction carried out under a normal air atmosphere. The study emphasizes the impact of carbon content (C/O ratios of 1 and 2) and heating temperature on the properties of the resulting ceramics. Morphology, chemical composition, phase analysis, and basic physical and mechanical properties are examined. This study presents a novel approach to industrial waste management by producing magnesia–hercynite ceramics through a single-step blending process followed by firing.

2. Materials and Methods

2.1. Materials Preparation

The magnesia powder used in the present study was supplied by Thai Poly Chemicals Co., Ltd., Samut Sakhon, Thailand. Aluminum dross (AD) in the form of fine powder was collected from Top Five Manufactory Co., Ltd., an aluminum dross melting company, Chachoengsao, Thailand. The AD was generated from the secondary dross melting process, primarily containing alumina as its major component. Mill scale (MS) was provided by UMC Metal Co., Ltd., an electric arc furnace steel mill, Chonburi, Thailand. The MS in the form of small flakes was ground into powder using a ball mill. The raw material was sifted through a sieve to isolate particles with sizes less than 180 µm. An X-ray fluorescence spectrometer (XRF) was employed to determine the chemical compositions of the magnesia, AD, and MS, as shown in

Table 1. Composition of the magnesia powder.

Oxides (wt%)
MgO	SiO2	CaO	Fe2O3
92.87	3.62	2.56	0.95

,

Table 2. Composition of the AD powder [24].

Oxides (wt%)
Al2O3	SiO2	Fe2O3	CaO	K2O	MgO	MnO	Na2O	SO3	CuO	TiO2	ZnO	Others
69.94	5.01	0.54	1.0	0.76	4.91	0.15	10.65	2.46	0.37	0.17	0.25	3.79

and

Table 3. Composition of the MS powder [24].

Oxides (wt%)
Fe2O3	SiO2	Al2O3	CaO	SO3	TiO2	K2O	P2O5	other
93.66	1.42	0.82	0.17	0.08	0.04	0.02	0.04	3.75

, respectively. The XRF used in the present study was the Rigaku ZSX Primus, Rigaku, Tokyo, Japan. The analysis was conducted using sample powder, scanning the elemental range from boron (B) to uranium (U).

The AD and MS were subsequently mixed with graphite powder, following C/O molar ratios of 1 and 2, and then mixed with the magnesia powder at 60–70 wt%, as shown in

Table 4. Composition of the blends at C/O = 1 (C60–C80) and C/O = 2 (D60–D80).

Blends	AD (wt%)	MS (wt%)	Graphite (wt%)	MgO (wt%)	Total (wt%)
C60	21.00	15.68	3.32	60	100
C70	15.75	11.76	2.49	70	100
C80	10.50	7.84	1.66	80	100
D60	19.40	14.48	6.12	60	100
D70	14.55	10.86	4.59	70	100
D80	9.70	7.24	3.06	80	100

. The blends (C60–C80 and D60–D80) were homogeneously mixed in a rolling mill for 30 min. C represents the total moles of carbon in the graphite, while O corresponds to the total moles of oxygen derived from Fe₂O₃ in the MS. The graphite was supplied by Kanto Chemical Co., Tokyo, Japan (Cat. No. 17046-02).

2.2. Material Synthesis

To form cylindrical pellets, approximately 5 g of the raw blends of C60–C80 and D60–D80 were each compressed in a cylindrical mold under a load of 10 kN for 2 min using a hydraulic press. The resulting green pellets had dimensions of approximately 9.76 mm in height and 13.21 mm in diameter, as shown in Figure 1. An overview of the experimental procedure is provided in Figure 2. The produced green pellets were dried in a hot air oven at 90 °C for 48 h to remove moisture. The dried pellets were then placed into alumina refractory crucibles, capable of withstanding temperatures up to 1800 °C, and transferred to a horizontal tube furnace at 1550 °C and 1650 °C for 1 h under normal air atmosphere. To prevent thermal shock, the crucibles were initially held in the cold zone of the furnace for 5 min before being moved to the hot zone. After 1 h, the heated pellets were quenched and prepared for further analysis. The pellets with C/O ratios of 1 (C60–C80) and 2 (D60–D80), after being heated at 1550 °C for 1 h, are presented in Figure 3. Compared with the raw pellets in Figure 1, the resulting ceramic pellets turned brown after the heating process. Based on visual observation, the color changed slightly from dark to light brown as the magnesia content increased from 60 wt% to 80 wt% and became slightly darker with increasing carbon content.

2.3. Formulation of the Reacting Raw Materials

In the magnesia–AD–MS–graphite system, the synthesis of magnesia–hercynite-based material can occur via the formation of hercynite and the reaction with MgO in the magnesia powder. Above 1550 °C, the formation of hercynite can occur through the carbothermic reduction of Fe₂O₃ in the system by solid carbon in graphite, as shown in Equations (1)–(3). The produced CO can further act as a reducing agent for Fe₂O₃, as shown in Equations (4)–(6). The CO₂ produced can react with solid carbon in the pellet to regenerate CO within the system, following the Boudouard reaction, as shown in Equation (7). The FeO produced can directly interact with Al₂O₃ to form hercynite, as represented by Equation (8), with a standard Gibbs free energy of ∆G° = −71,086 + 11.89∙T J∙mol⁻¹ [5,6]. Metallic Fe, derived from Equations (3) and (5), can be oxidized by oxygen to reform FeO. This FeO can then interact with Al₂O₃ to produce hercynite as the final product, as described in Equation (9), which has a standard Gibbs free energy of ∆G° = −328,348 + 82.004∙T J∙mol⁻¹ [5,6]. At 1550 °C, the ΔG° values for Equations (3), (6), (8), and (9) are −124.46, −271.71, −52.67, and −201.24 kJ·mol⁻¹, respectively, and they become even more negative at 1650 °C, suggesting that these reactions can occur at this experimental temperature. MgO in the magnesia powder can interact with FeO–Al₂O₃ and other impurity oxides in the system to form a magnesia–hercynite-based material [23], which needs to be investigated.

3Fe₂O₃ + C → 2Fe₃O₄ + CO

(1)

Fe₃O₄ + C → 3FeO + CO

(2)

FeO + C → Fe + CO

(3)

3Fe₂O₃ + CO → 2Fe₃O₄ + CO₂

(4)

Fe₃O₄ + CO → 3FeO + CO₂

(5)

FeO + CO → Fe + CO₂

(6)

CO₂ + C → 2CO

(7)

FeO (l) + Al₂O₃ (s) → FeAl₂O₄ (s)

(8)

$$\mathrm{Fe} (\mathrm{l})+{\mathrm{Al}}_2{\mathrm{O}}_3 (\mathrm{s})+{\displaystyle \frac{1}{2}}{\mathrm{O}}_2(\mathrm{g})\to \mathrm{Fe}{\mathrm{Al}}_2{\mathrm{O}}_4(\mathrm{s})$$

(9)

2.4. Material Analysis

X-ray diffraction (XRD) analysis of the samples was carried out using a Bruker D2 PHASER with Cu Kα (k = 1.54184 Å) radiation, Bruker, Bremen, Germany. XRD scans were carried out over the range (2theta angle) 10° to 80° at a step size of 0.05°/step. Phase identification was conducted using JADE 9.7.0 software and the ICDD 2022 database.

Microstructure examination and elemental analysis were conducted using scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS) with the JSM-7800 F instrument from JEOL, Akishima, Japan. The cold crushing strength of the heated pellets was measured using an H50KS universal testing machine obtained from Tinius Olsen, Philadelphia, PA, USA.

3. Results and Discussion

3.1. Effect of Carbon

In this section, the effect of carbon content in the magnesia–AD-MS–graphite system (C/O = 1 and 2) on the synthesis of magnesia–hercynite-based refractory was investigated.

The heated pellets were cross-sectioned and prepared for microscopic examination. Figure 4 and Figure 5 present the SEM micrographs of the cross-sectioned pellets and their corresponding EDS spectra for pellets with C/O ratios of 1 and 2, respectively. Numerous pores can be observed in the bulk ceramic in both cases. At a magnification of ×100, visual observation reveals that the porosity of the pellets decreases as the magnesia concentration in the blends increases from 60 to 80 wt%. At a magnification of ×1500, the ceramic morphology exhibits a rough surface for the pellets with a C/O ratio of 1 (Figure 4), while a smoother surface is observed as the carbon concentration in the blends increased to a C/O ratio of 2 (Figure 5). EDS analysis showed that the ceramic is primarily composed of Fe-Al-Mg-O-C in all cases. This suggests that the major components of the resulting ceramic are MgO, FeO, and Al₂O₃. An increase in the magnesia concentration in the raw blends from 60 to 80 wt% leads to an increase in the MgO content and a decrease in the FeO and Al₂O₃ contents in the resulting ceramic. However, the MgO content in the resulting ceramic remains lower than the initial MgO content in the raw blends. Increasing the carbon content in the system from a C/O ratio of 1 to 2 resulted in no significant difference in the MgO-FeO-Al₂O₃ contents according to the EDS results. The EDS analysis revealed that the carbon content in the resulting ceramics was lower than the initial carbon input in the raw blends. This reduction is attributed to the consumption of carbon during the carbothermic reduction of Fe₂O₃, as shown in Equations (1)–(3), and material loss during the combustion process.

Figure 6 presents the SEM images and EDS elemental distribution of the cross-sectioned pellets for blends C80 and D80 after heating at 1550 °C for 1 h. The images reveal that Mg forms equiaxed grains within the ceramic structure (green), while Al is densely packed in the areas between these grains (red). This occurs across the entire ceramic sample. The distribution of Fe, O, and C atoms is observed across the cross-section of the ceramic. This suggests that the resulting ceramic structure is primarily composed of equiaxed MgO grains (green region), with an FeO-Al₂O₃ phase (hercynite) at the grain boundary (red region). It is also observed that the carbon content in the raw blend, particularly the C/O ratios of 1 and 2, significantly affects the distribution of MgO and Al₂O₃ within the ceramic texture. In the pellet with a C/O ratio of 1, MgO and Al₂O₃ are densely packed in the equiaxed grain and grain boundary regions and also distributed throughout the cross-sectioned area. Conversely, in the pellet with a C/O ratio of 2, MgO and Al₂O₃ are primarily concentrated in the equiaxed grain and grain boundary regions, respectively. This suggests that the formation of MgO-FeO-Al2O4 spinel can occur and be distributed throughout the bulk ceramic. Additional XRF and XRD investigations were carried out for detailed composition analysis and phase identification.

Some of the resulting ceramics were ground into powder for XRF composition analysis, with the results presented in

Table 5. XRF analysis of the pellets at C/O = 1 (C60-C80) and C/O = 2 (D60-D80) after being heated at 1550 °C for 1 h.

Pellet	Oxides (wt%)
	Fe2O3	Al2O3	MgO	SiO2	CaO	Other
C60	25.52	12.50	51.16	5.93	3.44	1.45
C70	20.88	11.57	55.59	6.14	4.52	1.30
C80	15.29	9.04	64.60	5.37	4.88	0.82
D60	33.28	9.35	43.10	6.58	6.16	1.53
D70	27.81	11.30	49.85	6.15	3.71	1.18
D80	10.04	7.70	69.39	7.32	4.60	0.95

. In addition to the EDS results, the XRF analysis revealed that the major components of the resulting ceramic are MgO, Fe₂O₃, Al₂O₃, SiO₂, and CaO. The FeO content in the ceramic is represented by the Fe₂O₃ content in the XRF results (FeO = 0.89 × Fe₂O₃). Similar to the EDS results, the MgO content in the ceramic increases with an increase in the magnesia concentration in the raw blends, while the FeO and Al₂O₃ contents decrease. Increasing the carbon content in the system (C60–C80 and D60–D80) results in no significant difference in the chemical composition of the resulting ceramic. Compared with the initial MgO content in the raw blends (60–80 wt%), the MgO content in the resulting ceramics decreases by approximately 20% in samples C60–C80 (51.2–64.6 wt%), and by around 10–30% in samples D60–D80 (43.1–69.4 wt%). This reduction is expected because of the consumption of MgO in the formation of other ceramic compounds and material loss during the combustion process.

XRD phase identification of the resulting ceramic was conducted, as shown in Figure 7. It was found that MgO, Fe₂O₃, Al₂O₃, SiO₂, and CaO in the ceramic interacted to form various ceramic compounds. The detected phases included periclase (MgO), hercynite (FeAl₂O₄), magnesium iron aluminum oxide (MgFeAlO₄), calcium magnesium silicate (CaMg(SiO₄)), merwinite (Ca₃Mg(SiO₄)₂), and carbon. The presence of FeAl₂O₄ indicates an interaction between FeO and Al₂O₃, as described in Equations (8) and (9), while the formation of the other compounds suggests interactions between MgO and the other oxides (Fe₂O₃, Al₂O₃, SiO₂, and CaO). The Rietveld refinement method was conducted using JADE 9.7.0 software and the ICDD 2022 database for the quantitative phase analysis of the resulting ceramic in all cases. Figure 8 presents an example of the XRD Rietveld refinement for the pellets of blend C60 (C/O = 1) after heating at 1550 °C for 1 h. The ICDD PDF card numbers of the detected phases are MgO: #98-000-0349, FeAl₂O₄: #98-000-0242, MgFeAlO₄: #04-018-0090, CaMg(SiO₄): #01-084-1319, Ca₃Mg(SiO₄)₂: #98-000-0304, and C: #98-000-0231.

The quantitative phase composition of the pellets with C/O ratios of 1 (C60–C80) and 2 (D60–D80) after heating at 1550 °C for 1 h is shown in

Table 6. Rietveld refinement quantitative phase composition of the pellets at C/O = 1 (C60–C80) and C/O = 2 (D60–D80) after being heated at 1550 °C for 1 h.

Pellet	Oxides (wt%)
	MgO	FeAl2O4	MgFeAlO4	CaMg(SiO4)	Ca3Mg(SiO4)2	C
C60	43.2	14.5	30.5	7.5	-	4.3
C70	60.7	9.0	24.1	6.2	-	-
C80	73.3	5.7	14.4	2.8	3.8	-
D60	54.4	12.0	25.6	8	-	-
D70	47.3	15.2	28.8	8.2	-	-
D80	76.2	5.2	9.6	2.1	6.9	-

. The major phase in the resulting ceramic is MgO, with contents ranging from 43.2 to 73.3 wt% for the C60–C80 pellets and from 54.4 to 76.2 wt% for the D60–D80 pellets. Increasing the MgO content in the raw blends leads to an increase in the MgO content in the resulting ceramic. As the MgO content increases, the amounts of FeAl₂O₄ and MgFeAlO₄ tend to decrease. For the pellets with a C/O ratio of 1, C60 contains a moderate amount of MgO, along with significant quantities of MgFeAlO₄ and FeAl₂O₄. A small quantity of CaMg(SiO₄) is present, and a trace of carbon is detected. In C70, the MgO content is higher than in C60, while the levels of FeAl₂O₄ and MgFeAlO₄ are reduced. A small amount of CaMg(SiO₄) is also present. C80, which has the highest MgO content among the pellets with a C/O ratio of 1, shows further reductions in FeAl₂O₄ and MgFeAlO₄. Ca₃Mg(SiO₄)₂ is detected, along with a small amount of CaMg(SiO₄). For the pellets with a C/O ratio of 2, D60 exhibits a relatively high MgO content, along with considerable amounts of MgFeAlO₄ and FeAl₂O₄. It also has the highest CaMg(SiO₄) content among all the pellets, with no Ca₃Mg(SiO₄)₂ detected. In D70, the MgO content is lower than in D60, but it has the highest FeAl₂O₄ and MgFeAlO₄ contents among the pellets with a C/O ratio of 2. The CaMg(SiO₄) content remains similar to that in D60, with no Ca₃Mg(SiO₄)₂ detected. D80 has the highest MgO content among all the pellets, with significantly reduced levels of FeAl₂O₄ and MgFeAlO₄. Ca₃Mg(SiO₄)₂ is present at a higher level than in C80, while CaMg(SiO₄) is present in a small amount.

The basic physical properties, including bulk density and apparent porosity, as well as mechanical properties of the resulting ceramic with C/O ratios of 1 and 2 after heating at 1550 °C for 1 h, are shown in Figure 9. It was found that varying the carbon content (C/O ratios of 1 and 2) in the raw blends has an insignificant effect on the bulk density of the resulting ceramic, which ranges between 2.73 and 2.99 g/cm³. Increasing the amount of magnesia powder in the raw blends also shows no significant impact on the bulk density of the resulting ceramic. The apparent porosity of the resulting ceramic tends to decrease with increasing amounts of magnesia powder in the raw blends. For pellets with a C/O ratio of 1, the apparent porosity decreases from 8.66% to 4.11% vol, while for pellets with a C/O ratio of 2, it decreases from 12.68% to 3.87% vol. For blends containing 60 wt% MgO, the pellet with higher carbon content (D60) exhibits higher apparent porosity (12.68% vol) compared with the pellet with a lower carbon content (C60), which has an apparent porosity of 8.66% vol. However, for blends with higher MgO content (70 and 80 wt%), the trend is reversed, with pellets C70 and C80 showing higher apparent porosity than D70 and D80. This observation is also supported by the visual evidence of pore numbers from the SEM images shown in Figure 4 and Figure 5.

In terms of mechanical properties, the cold crushing strength of the resulting ceramic shows a clear trend of increasing with both higher MgO and carbon concentrations in the raw blends. This relationship is evident across different C/O ratios. For the pellets with a C/O ratio of 1, specifically those labeled C60 through C80, the cold crushing strength demonstrates a noticeable increase. The strength ranges from 43.77 MPa for the C60 pellet to 58.97 MPa for the C80 pellet. This trend highlights how the addition of MgO contributes to enhancing the mechanical robustness of the ceramic. In contrast, for the pellets with a C/O ratio of 2, which include D60 through D80, the cold crushing strength is generally higher. These pellets exhibit a range from 38.96 MPa for D60 to an impressive 78.61 MPa for D80. This indicates that the combination of increased MgO content and higher carbon concentrations results in a more substantial increase in crushing strength compared with the pellets with a C/O ratio of 1. The data also reveal that pellets with the highest MgO content, namely, C80 and D80, provide the highest cold crushing strength compared with those with lower MgO content. Specifically, C80, with its high MgO concentration, achieves a strength of 58.97 MPa, whereas D80, which not only has high MgO content but also a higher carbon concentration, reaches a maximum strength of 78.61 MPa.

The increase in carbon content from a C/O ratio of 1 to 2 is observed to enhance the cold crushing strength of the ceramic significantly. This is particularly evident when comparing the strength of 58.97 MPa in C80 (with a C/O ratio of 1) to 78.61 MPa in D80 (with a C/O ratio of 2). This improvement highlights the combined influence of higher MgO levels and increased carbon content on the mechanical properties of the ceramic material. The higher carbon content present in the raw blends provides more carbon for the carbothermic reduction of Fe₂O₃ present in the mill scale, available in both solid form and CO gas. This increased availability of carbon also facilitates other reactions between the oxides in the blends, leading to the formation of various ceramic compounds. As a result, these changes contribute to a reduction in bulk density (2.99 to 2.94 g/cm³) and apparent porosity (4.11% to 3.87% vol), which, in turn, enhances the cold crushing strength of the ceramic. This relationship suggests that the optimized combination of MgO and carbon not only affects the microstructure but also plays a role in determining the overall mechanical strength of the ceramic.

3.2. Effect of Temperature

In this section, the effect of heating temperatures of 1550 °C and 1650 °C on the magnesia–AD-MS–graphite system for synthesizing magnesia–hercynite-based refractory is investigated. The carbon content in the raw blends was fixed at a C/O ratio of 1. Pellets with C/O ratios of 1 (C60–C80), after being heated at 1550 °C and 1650 °C for 1 h, are shown in Figure 10. It can be visually observed that the color of the pellets changes slightly from light brown to a deeper, stronger brown as the heating temperature increases.

Figure 11 presents SEM micrographs of the cross-sectioned pellets and their corresponding EDS spectra for pellets with a C/O ratio of 1 after being heated at 1650 °C for 1 h. A number of pores can be observed throughout the bulk ceramic in all samples (C60–C80). However, compared with the results shown in Figure 4, the porosity appears to have decreased visually, and the surface of the ceramic is noticeably smoother. The EDS analysis indicated that the major components in the resulting ceramic are MgO, FeO, and Al₂O₃, which suggests the formation of related compounds within the ceramic, such as FeAl₂O₄ and MgFeAlO₄. The increase in heating temperature did not significantly affect the EDS analysis results in terms of composition. However, further XRF and XRD analyses are essential for both quantitative and qualitative assessments.

Figure 12 presents the SEM images and EDS elemental distribution of the cross-sectioned C80 pellets after heating at 1650 °C for 1 h. It is observed that Mg (shown in green) forms equiaxed grains within the ceramic structure, while Al (shown in red) is densely packed in areas separate from these equiaxed grains. The distribution of iron (Fe), oxygen (O), and carbon (C) atoms is also observed across the ceramic’s cross-section. This suggests that the resulting ceramic structure is primarily composed of equiaxed MgO grains (green regions) and an FeO-Al₂O₃ phase (red regions). Compared with the structure observed at 1550 °C in Figure 6a, the red regions corresponding to the FeO-Al₂O₃ phase increase and are more clearly separated from the green MgO regions. The formation of MgFeAlO₄, CaMg(SiO₄), and Ca₃Mg(SiO₄)₂ is expected to distribute throughout the bulk ceramic and mainly at the red region or boundary of the MgO grain.

Table 7. XRF analysis of the pellets at C/O = 1 (C60–C80) after being heated at 1650 °C for 1 h.

Pellet	Oxides (wt%)
	Fe2O3	Al2O3	MgO	SiO2	CaO	Other
C60	26.33	13.67	46.21	8.12	4.12	1.55
C70	20.76	12.82	51.06	8.58	5.22	1.56
C80	14.54	10.89	60.32	7.99	5.33	0.93

presents the XRF analysis of the pellets with a C/O ratio of 1 (C60–C80) after heating at 1650 °C for 1 h. The resulting ceramic is composed of MgO, Fe₂O₃, Al₂O₃, SiO₂, and CaO, which is not significantly different from the composition observed at 1550 °C, as shown in Table 5. The MgO content in the resulting ceramics of C60 to C80 increases from 46.21 wt% to 60.32 wt%, while the Fe₂O₃ and Al₂O₃ contents decrease from 26.33 wt% to 14.54 wt% and from 13.67 wt% to 10.89 wt%, respectively.

The XRD analysis in Figure 13 reveals that the phases present in the resulting ceramic include MgO, FeAl₂O₄, MgFeAlO₄, CaMg(SiO₄), Ca₃Mg(SiO₄)₂, and carbon, with variations depending on the magnesia content in the raw blends. The Rietveld refinement results for the quantitative phase composition of the pellets with a C/O ratio of 1 (C60–C80) after heating at 1650 °C for 1 h are shown in

Table 8. Rietveld refinement quantitative phase composition of the pellets at C/O = 1 (C60–C80) after being heated at 1650 °C for 1 h.

Pellet	Oxides (wt%)
	MgO	FeAl2O4	MgFeAlO4	CaMg(SiO4)	Ca3Mg(SiO4)2	C
C60	54.6	8.9	29.0	7.4	-	-
C70	58.1	14.7	19.9	7.3	-	-
C80	77.9	7.8	12.9	3.0	3.6	0.7

. As the MgO content increases from C60 to C80, there are noticeable changes in the composition of other oxides. In C60, the pellet contains 54.6 wt% MgO, with significant amounts of MgFeAlO₄ (29.0 wt%) and FeAl₂O₄ (8.9 wt%). The pellet also includes 7.4 wt% CaMg(SiO₄), with no detectable amounts of Ca₃Mg(SiO₄)₂ or carbon. In C70, the MgO content rises to 58.1 wt% and FeAl₂O₄ increases to 14.7 wt%, while MgFeAlO₄ decreases to 19.9 wt%. The CaMg(SiO₄) content remains relatively stable at 7.3 wt%, with no detection of Ca₃Mg(SiO₄)₂ or carbon. C80, which has the highest MgO content at 77.9 wt%, shows a significant decrease in FeAl₂O₄ and MgFeAlO₄ to 7.8 wt% and 12.9 wt%, respectively. The CaMg(SiO₄) content decreases to 3.0 wt%, Ca₃Mg(SiO₄)₂ appears at 3.6 wt%, and a small amount of carbon (0.7 wt%) is also detected. Compared with the data in Table 6, increasing the heating temperature from 1550 °C to 1650 °C results in an increase in the amounts of MgO and MgFeAlO₄ phases. The quantities of other phases show only slight changes.

Figure 14 illustrates the bulk density, apparent porosity, and cold crushing strength of the pellets with a C/O ratio of 1 (C60–C80) after heating at 1550 °C and 1650 °C for 1 h. Increasing the heating temperature from 1550 °C to 1650 °C results in a slight reduction in bulk density. At 1550 °C, the bulk density ranges from 2.85 to 2.99 g/cm³, while at 1650 °C, it ranges from 2.78 to 2.93 g/cm³. The apparent porosity shows a significant decrease with increasing temperature. At 1550 °C, the apparent porosity ranges from 8.66% to 4.11% vol, whereas at 1650 °C, it reduces to a range of 4.31% to 1.58% vol. A lower apparent porosity indicates fewer void spaces or pores within the material, as supported by the SEM images in Figure 11. Comparing the pellets heated at 1550 °C and 1650 °C, although the measured densities are nearly the same, the apparent porosity differs. This may be attributed to the types of pores present in the bulk ceramics, specifically open and closed pores. Heating at the higher temperature of 1650 °C may promote the formation of new phases that can fill the pores, thereby reducing porosity. The reduction in apparent porosity of approximately 50% suggests that the ceramic becomes denser and less porous at the higher temperature, reflecting improved material compactness and reduced void content. These observations indicate that heating to 1650 °C enhances the densification of the ceramic material, leading to a more compact structure with lower porosity. As a result, the mechanical properties of the ceramic significantly improve with the increase in heating temperature. The cold crushing strength of the ceramic heated at 1550 °C ranges from 43.77 to 58.97 MPa. However, when the temperature is increased to 1650 °C, the cold crushing strength nearly doubles, reaching values between 76.79 and 95.67 MPa.

3.3. Evaluation of the Magnesia–Hercynite Ceramic Produced from Commercial and Waste Resources

Based on the experimental results, the ceramic pellets of C80 and D80 exhibit the optimal conditions among all tested samples, demonstrating favorable bulk density, low apparent porosity, and high cold crushing strength. These pellets were selected for comparison with commercial magnesia–hercynite bricks, as detailed in

Table 9. Comparison of magnesia–hercynite ceramic produced from commercial and waste resources.

Refractory Sample	Commercial Resources		Present Study
	RFC#1 [2]	RFC#2 [4]	C80 at 1550 °C	D80 at 1550 °C	C80 at 1650 °C
Raw materials
Magnesia	China	India	Thailand magnesia
Hercynite	Commercial electro-fused		Mill scale and aluminum dross combustion
Chemical composition (wt%)
MgO	85.0	87.8	64.6	69.93	60.32
Fe2O3	3.8	4.8	15.29	10.04	14.54
Al2O3	3.4	4.89	9.04	7.7	10.89
CaO	0.7	1.25	4.88	4.6	5.33
SiO2	0.3	0.6	5.37	7.32	7.99
Basic physical properties
Bulk density (g/cm3)	3.06	2.97	2.99	2.94	2.81
Apparent porosity (vol%)	14.0	16.5	4.11	3.87	1.58
Cold crushing strength (MPa)	70	57.3	58.97	78.61	95.67

. The following commercial magnesia–hercynite bricks were used for comparison: one from China (RFC#1) and one from India (RFC#2). The MgO source for both RFC#1 and RFC#2 is commercial magnesia. The hercynite in these samples was produced using the electro-fused method with commercial alumina and iron sources. The chemical compositions of RFC#1 and RFC#2 are quite similar. Both bricks have a high MgO content, with RFC#1 containing 85 wt% and RFC#2 containing 87.8 wt%. The major components are followed by Fe₂O₃, which ranges from 3.8 to 4.8 wt%, and Al₂O₃, which ranges from 3.4 to 4.9 wt%. Both samples also include small amounts of CaO and SiO₂. Regarding their basic physical properties, RFC#1 and RFC#2 have nearly identical bulk density and apparent porosity. However, RFC#1 has a cold crushing strength of 70 MPa, whereas RFC#2 has a lower strength of 57.3 MPa. Compared with the commercial magnesia–hercynite bricks, the ceramics produced in the present study have a lower MgO content, ranging from approximately 64 to 70 wt%. In contrast, the Fe₂O₃ and Al₂O₃ components are more than twice as high as those in the commercial ceramics, ranging from approximately 10 to 15 wt% and 8 to 11 wt%, respectively. Because of the waste materials used, the CaO and SiO₂ contents are also significantly higher in the ceramics from the present study. The ceramics produced in the present study exhibit a bulk density similar to that of RFC#1 and RFC#2, ranging from 2.8 to 3 g/cm³, but with significantly lower apparent porosity. The cold crushing strength is also comparable, ranging from approximately 59 to 95 MPa.

Ding et al. [22] investigated the effect of hercynite content on the properties of magnesia–spinel composite refractories sintered in different atmospheres. It was reported that the addition of 2–3 wt% FeAl2O4 to magnesia–spinel composite refractories improved sintering and enhanced thermal conductivity. Among the tested specimens, the one with 2 wt% FeAl₂O₄ sintered in an argon atmosphere exhibited the highest thermal conductivity and optimal overall properties. The study also indicated that Fe²⁺ was more effective than Fe³⁺ in enhancing the sintering process of MgO-MgAl₂O₄. Jiang et al. [23] studied the morphology characterization of periclase–hercynite refractories produced through reaction sintering. It was concluded that during the reaction sintering of periclase–hercynite bricks at 1600 °C, cation diffusion between periclase and hercynite crystals significantly influenced microstructure development. Fe²⁺, Fe³⁺, and Al³⁺ ions from hercynite migrated and reacted with periclase to form spinels with varying Fe/Al ratios. Simultaneously, Mg²⁺ ions from periclase (MgO) migrated into hercynite, leading to spinel formation with a lower Fe/Al ratio. This cation exchange enhanced the sintering process and resulted in a microporous structure in the periclase–hercynite brick.

In the present study, impurities in the oxides present in the aluminum dross (AD) and mill scale (MS) may account for the variations in chemical composition compared with commercial bricks. These impurities are expected to influence the strength and apparent porosity of the resulting ceramic. In the FeO-Al₂O₃-MgO ternary system, the formation of FeAl₂O₄ and MgFeAlO₄ phases was observed in the resulting ceramic, with MgO as the primary component. This occurs because of the migration of Mg²⁺ ions from MgO into hercynite, as well as the migration of Fe²⁺, Fe³⁺, and Al³⁺ ions from hercynite, which react with MgO [23]. Because of the presence of high amounts of impurity oxides, the combination of MgO, SiO₂, and CaO can form a liquid phase in the ternary SiO₂-CaO-MgO system, which has a melting temperature of 1500 °C [25]. According to the SiO₂-CaO-MgO phase diagram [25], this process results in the formation of CaMg(SiO₄) and Ca₃Mg(SiO₄)₂ phases because of the heating temperatures of 1550 °C and 1650 °C used in the present study [23]. The formation of the merwinite phase (Ca₃Mg(SiO₄)₂) is observed only in samples with high magnesia content (C80 and D80). The migration of Fe²⁺, Fe³⁺, and Al³⁺ ions from hercynite into the liquid phase of the SiO₂-CaO-MgO system transforms the boundary liquid phase from a SiO₂-CaO-MgO system to a SiO₂-CaO-MgO-Fe₂O₃-Al₂O₃ system. This change reduces both the melting temperature and the viscosity of the boundary liquid [26]. These are expected to influence the microstructure of the resulting ceramic, impacting both the cold crushing strength and apparent porosity. As observed from the SEM images and elemental distribution (Figure 6 and Figure 12), the resulting ceramic consists of equiaxed MgO grains (green regions) and an FeO-Al₂O₃ phase (red regions) at the boundaries. The formation of MgFeAlO₄, CaMg(SiO₄), and Ca₃Mg(SiO₄)₂ is anticipated to occur throughout the bulk ceramic but is primarily concentrated at the red regions or boundaries of the MgO grains.

4. Conclusions

In the present study, the synthesis of magnesia–hercynite-based refractories was conducted using various blends of magnesia powder, aluminum dross (AD), mill scale (MS), and graphite. The effects of carbon concentration and heating temperatures were then investigated. The experimental results can be summarized as follows:

• Magnesia–hercynite-based refractories can be successfully synthesized from MgO, MS, AD, and graphite via reactions at 1550 °C and 1650 °C for 1 h in a normal air atmosphere. The single-step blending of raw materials was employed for this process.
• The combination of blend samples with high magnesia content (C80 and D80) and high levels of impurity oxides (CaO and SiO₂) leads to variations in the chemical composition and the formation of distinct phases in the resulting ceramic. The phases present in the resulting ceramic include MgO, FeAl₂O₄, MgFeAlO₄, CaMg(SiO₄), and Ca₃Mg(SiO₄)₂. These phases, in turn, affect the microstructure, as well as the physical and mechanical properties of the produced magnesia–hercynite-based refractory.
• The produced magnesia–hercynite-based refractories consist of equiaxed MgO grains and an FeO-Al₂O₃ spinel phase at the boundaries. The formation of MgFeAlO₄, CaMg(SiO₄), and Ca₃Mg(SiO₄)₂ is anticipated to occur throughout the bulk ceramic, but it is primarily concentrated at the red regions or boundaries of the MgO grains.
• Varying the carbon content (C/O ratios of 1 and 2) has a minimal impact on the bulk density and apparent porosity of the resulting ceramic. Increasing the magnesia powder content (60–80 wt%) does not significantly affect bulk density but tends to reduce apparent porosity. A higher MgO content combined with increased carbon concentration (C/O = 2) leads to a more notable increase in crushing strength compared with a C/O ratio of 1. At 1550 °C, the pellets with the highest MgO content (C80 and D80) exhibit the greatest cold crushing strength.
• Increasing the heating temperature from 1550 °C to 1650 °C results in a slight reduction in bulk density and a significant decrease in apparent porosity, enhancing the ceramic’s densification and compactness. This leads to a notable improvement in mechanical properties, with cold crushing strength nearly doubling from 43.77–58.97 MPa at 1550 °C to 76.79–95.67 MPa at 1650 °C.
• Blends with high magnesia content (C80 and D80) show optimal conditions for synthesizing magnesia–hercynite-based refractories from the magnesia–dross–scale–graphite system, exhibiting physical and mechanical properties comparable to commercial magnesia–hercynite bricks. However, the quantity and type of phases in the ceramic differ from those in commercial bricks, indicating that further investigation of the ceramic’s large-scale and thermal properties is necessary for potential application in industrial rotary kilns.