Hot Modification of Silicomanganese Slag in Its Crystallization and Viscosity Properties for Preparation of Cast Stone

Abstract

The direct hot modification and subsequent preparation of qualified building materials from molten slag has gained significant attention at present due to its characteristics of saving energy and reducing CO₂ emissions. Molten silicomanganese slag, discharged at 1500–1600 °C with high content of SiO₂ and Al₂O₃ (above 50 mass%), was suitable for the preparation of casting stone. To ensure a qualified casting stone, the study focused on improving the crystallization properties and fluidity of molten silicomanganese slag by modifying of its composition, crystallization, structure, and viscosity. The raw slag and two modified slags were compared, and the physical properties of their final cast stone were discussed. The results showed that after being modified by addition of 10 mass% chromite and serpentine or 20 mass% ferrochrome slag into the silicomanganese slag, both the crystallization ability and fluidity of the molten slag were improved simultaneously. Augite and spinel precipitated in the modified slag, compared with glass phase in the raw slag. The precipitation of spinel, on the one hand, acted as a nucleation agent, dynamically promoting the formation of augite, and on the other hand, increased the proportion of SiO₂ and its polymerization of [SiO₄] structural units in the residual liquid slag, further promoting the generation of augite in the composition and structure. The gradual precipitation of crystals effectively mitigated sudden viscosity fluctuations resulting from crystallization, contributing to a smooth casting process for molten slag. Both cast stones from the modified slag exhibited qualified physical properties, compared with the broken glass from the raw slag. This indicated the feasibility of low-cost modification during the discharging process of molten silicomanganese slag by blending 10 mass% cold modifiers or 20 mass% molten ferrochrome slag into it.

1. Introduction

Currently, most slag discharged at temperatures between 1200 °C and 1600 °C is cooled using air or water, leading to significant waste heat loss and low slag utilization rates. This issue has garnered increasing attention in the fields of energy recovery, resource recycling, and environmental protection [1,2,3,4,5].

The direct utilization of molten slag for the preparation of building materials, such as glass-ceramic and cast stone, has emerged as a prominent research focus [6,7,8,9,10,11]. This approach not only facilitates the resourceful utilization of slag but also maximizes the recovery of its thermal energy. Among these methods, the preparation of cast stone employs the petrurgic method. Unlike the two-step or one-step methods, which involve cooling the slag into a glass phase before nucleation and crystallization, the petrurgic method directly induces crystallization during the cooling process of the melt [12]. This process is analogous to the natural cooling and solidification of magma into lava. The petrurgic method has been traditionally used to manufacture cast stone pipes or plates by heating and melting basalt. Numerous studies have also applied this method to produce glass-ceramics or cast stone from blast-furnace slag [10,11,13], steel slag [8,14], copper slag [15,16], and silicomanganese [6,9], among other materials.

Since the petrurgic method does not include a dedicated nucleation step, the slags used must exhibit excellent inherent nucleation and crystallization properties. Current research in this area primarily focuses on enhancing the crystallization performance of slag-based cast stones through the addition of nucleation agents [6,9,17,18] and optimization of crystallization temperatures [6,9,17,19]. Additionally, the slags must demonstrate “long slag” characteristics during casting and shaping, meaning their viscosity decreases gradually and slowly as the temperature drops [20,21]. Excessively rapid crystallization can cause a sharp increase in viscosity, disrupting the casting process. Silicomanganese slags, which are produced in large volumes and discharged at high temperatures, present challenges for large-scale utilization [16,22,23]. However, their high silicon and aluminum content makes them suitable for the preparation of cast stone materials. Directly utilizing silicomanganese slags for cast stone materials not only enhances their value but also eliminates the need for remelting, contributing to energy conservation and emission reduction. This holds significant importance.

Direct air cooling of silicomanganese slag often results in an uneven microstructure and significant thermal stress, making it unsuitable for direct production of high-quality cast stone products. Therefore, conditioning treatment is necessary for silicomanganese slag. Previous studies have shown that chromium-containing modifiers can effectively regulate the crystallization properties of silicomanganese slag, facilitating the production of high-quality glass-ceramics or cast stone. For example, Deng et al. [21] successfully synthesized CaO-MgO-Al₂O₃-SiO₂ glass-ceramics with varying Cr₂O₃ concentrations, using blast furnace slag as the primary raw material through traditional smelting techniques. The resulting glass-ceramics consisted of dendritic diopside (primary crystalline phase) and bulk spinel (secondary crystalline phase), with crystallization activation energies ranging from 378 to 454 kJ/mol and a crystallization index increasing from 0.84 to 2.80. Similarly, Yang [6] used silicomanganese slag as the primary raw material, incorporating chrome-bearing modifier such as silica, iron scale, and chrome-iron slag, to produce pyroxene glass-ceramics that meet the performance criteria of natural granite stone using the petrurgic one-step method. The modified glass-ceramics exhibited significantly enhanced crystallization properties compared to the raw silicomanganese slag, with ferrochrome slag promoting the formation of granular or short rod-shaped pyroxene crystals with particle sizes ranging from 0.2 to 0.5 μm.

Jin et al. [24] investigated the key challenges of using blast furnace slag directly in glass production. They found that while high-temperature melts exhibit low viscosity and good fluidity, the rapid increase in viscosity during cooling leads to material brittleness and poor formability. By modifying the slag composition with SiO₂ and other oxides, they successfully extended the material properties and achieved temperature–viscosity curves typical of glass melts, overcoming the technical bottleneck of direct forming of blast furnace slag. Building on this, Wang et al. [25] further explored the melting process for preparing glass-ceramics from blast furnace slag and systematically studied the regulation mechanism of nucleating agents on process parameters. Their studies revealed that while 2 mass% Cr₂O₃ alone increased melt viscosity, requiring a pouring temperature of 1400 °C or higher to ensure fluidity, the introduction of 2 mass% Fe₂O₃ effectively reduced the melting point to 1325 °C, albeit with limited viscosity regulation. A combined addition of both nucleating agents (2 mass% each) achieved an optimal balance between melt fluidity and crystallization at an equilibrium temperature of 1375 °C. In summary, to ensure a smooth casting process, the modified slag must not only exhibit excellent crystallization properties but also maintain suitable viscosity characteristics to prevent rapid viscosity increases due to crystallization. However, research on the sensible heat modification of slag and the preparation of high-quality cast stone by adjusting crystallization and viscosity remains limited.

In this study, molten silicomanganese slag was used to melt modifiers through sensible heat, and cast stone was prepared using the petrurgic method. The primary objective was to investigate the changes and underlying mechanisms in slag crystallization and viscosity following the addition of modifiers, ensuring both the smooth pouring of the slag and the satisfactory performance of the cast stone. This research provides a theoretical foundation for the development of an economical, efficient, and environmentally friendly low-carbon technology utilizing sensible heat. Furthermore, it demonstrates promising application potential in the production of cast stone and glass-ceramic building materials.

2. Materials and Experiments

2.1. Raw Materials

The experimental study employed raw materials comprising silicomanganese slag, chromite, serpentine, and ferrochrome slag, all sourced from Inner Mongolia, China. The chemical composition and mineral phases of the quenched silicomanganese slag and modifiers are detailed in

Table 1. Chemical compositions of the raw material (mass%).

Raw Material	CaO	SiO2	Al2O3	MgO	MnO	Fe2O3	K2O	Cr2O3	BaO	Total
Silicomanganese slag	27.27	38.40	13.24	3.60	7.92	2.27	1.60	0.26	2.49	97.05
Chromite	1.04	8.30	10.25	9.84	0.56	26.40	0.06	40.50	—	96.95
Serpentine	1.53	43.29	2.01	39.31	0.22	11.05	0.06	0.73	—	98.20
Ferrochrome slag	4.90	28.80	23.20	26.10	0.60	4.23	0.12	9.73	—	97.68

and illustrated in Figure 1. The unmodified silicomanganese slag was designated as S1, while the modified slags were prepared according to specific mass ratios: silicomanganese slag/chromite/serpentine = 92:6:2 (designated as S2) and silicomanganese slag/ferrochrome slag = 80:20 (designated as S3). The chemical composition analysis results for these three samples are presented in

Table 2. Chemical composition of mixed silicomanganese slag (mass%).

Sample No.	CaO	SiO2	Al2O3	MgO	MnO	Fe2O3	K2O	Cr2O3	BaO	R (mass%CaO/mass%SiO2)
S1	27.27	38.40	13.24	3.60	7.92	2.27	1.60	0.26	2.49	0.71
S2	25.18	36.69	12.84	4.69	7.32	3.89	1.48	2.68	2.29	0.69
S3	22.80	36.48	15.23	8.10	6.46	2.66	1.30	2.15	1.99	0.63

.

2.2. Sample Preparation

Three distinct formulations of samples (with four replicates each) were placed into corundum crucibles (99% Al₂O₃) and heated to 1500 °C in a muffle furnace for 30 min to ensure complete melting of the slag. Subsequently, three crucibles containing different samples were removed from the furnace at temperatures of 1500 °C, 1300 °C, 1100 °C, and 1000 °C, respectively, after being held at each temperature for 30 min, followed by immediate quenching. The phase composition of the quenched slag at these temperatures was analyzed using X-ray diffraction (XRD).

XRD analysis revealed that all slag samples quenched at 1500 °C exhibited an amorphous structure, as illustrated in Figure 2. The XRD patterns of the quenched slag displayed a prominent broad diffraction peak in the range of 2θ = 20–40°, with minimal additional diffraction peaks. This observation confirms that the samples were predominantly composed of amorphous glass material, indicating that the raw materials and modifiers were thoroughly integrated, resulting in a chemically homogeneous composition.

The experiment was designed to simulate the mixing process of hot slag with modifiers, even though the cold slag was remelted after being mixed with the modifiers. The experimental procedure is outlined in Figure 3a, with the red dotted box emphasizing the mechanism of the petrurgic method. The temperature profile used in the experiment is shown in Figure 3b. Specifically, 500 g of dried slag samples were placed in a corundum crucible equipped with a corundum protective cover. The crucible was then positioned in a muffle furnace, heated to 1500 °C, and maintained at this temperature for 60 min. Afterward, the temperature was reduced to 1300 °C. The molten slag was then poured directly into a stainless-steel mold preheated to 700 °C. The mold was subsequently transferred to another muffle furnace and held at a crystallization temperature of 1000 °C for 60 min. Following this, the temperature was lowered to an annealing temperature of 700 °C, maintained for 60 min, and then allowed to cool within the furnace. This process ultimately yielded the silicon manganese slag cast stone sample.

2.3. Experimental Methods

2.3.1. XRF and XRD Measurement

The samples were crushed and ground, and screened with a 200-mesh screen. The chemical composition and minerals of raw materials were analyzed by X-ray fluorescence (XRF, Rigaku ZSX Primus, Tokyo, Japan) spectrometer and X-ray diffraction (XRD, Rigaku Smartlab 9 kW, Tokyo, Japan) spectrometer, respectively. The XRD test adopted a Cu target; the scanning speed was 10°/min, and scanning range was 10–90°.

2.3.2. Non-Isothermal DSC Measurement

The non-isothermal crystallization characteristics of silicomanganese slag were studied by differential scanning calorimetry (DSC, Netzsch STA449F3; Netzsch Instrument Inc., Waldkraiburg, Germany) under an argon (Ar) atmosphere at 60 mL min⁻¹. Approximately 50 mg of slag sample was placed into a platinum crucible with an inner diameter of 5 mm and a height of 5.5 mm. The slag was heated from room temperature to 1500 °C at 30 °C min⁻¹ and held for 1 min to eliminate air bubbles and homogenize the chemical composition and temperature. Subsequently, the temperature was reduced to 350 °C with cooling rates of 10 °C min⁻¹, 20 °C min⁻¹, and 30 °C min⁻¹. DSC data were automatically recorded during the heating and cooling cycles.

2.3.3. Viscosity Measurement

The viscosity of molten slag was measured using a rotary viscometer (Model DV2T-LV, Brookfield, WI, USA) employing the rotary cylinder method. The heating element of the blast furnace consisted of a U-shaped MoSi₂ rod, with temperature measurements conducted using a B-type (Pt-30% Rh: Pt-6% Rh) thermocouple. To prevent chemical reactions with the slag melt, molybdenum (Mo) material was utilized for the crucible, probe, and connecting rod in the experiments. Prior to each measurement, three standard silicone oils with viscosities of 0.0973 Pa s, 0.490 Pa s, and 0.985 Pa s were subjected to a 30 min equilibration period in a constant-temperature water tank set at 25 °C for viscosity calibration. Subsequently, approximately 250 g of slag was loaded into a Mo crucible shielded by a graphite crucible under Ar gas (99.99%, 0.5 NL min⁻¹). The slag was then heated to 1550 °C and maintained at this temperature for 30 min to ensure homogenization of both the slag’s temperature and chemical composition. Following this, a Mo rotor was introduced into the slag and rotated at 13 r min⁻¹ to measure the viscosity. Throughout the continuous cooling process, the slag was cooled at a rate of 5 °C/min, and viscosity measurements were conducted at intervals to monitor changes in viscosity as the slag cooled.

2.3.4. FTIR and XPS Spectra Measurement

The structural characteristics of the glass samples were determined using Fourier transform infrared (FTIR, Nicolet iS50, Thermo Fisher, St. Waltham, MA, USA) spectroscopy. About 3.0 mg of slag powders mixed with 200 mg KBr (reagent grade) was pressed into a thin section disk for FTIR measurement. The FTIR measurement was taken using a spectrophotometer equipped with a KBr detector, and the spectra were recorded in the range of 4000–400 cm⁻¹. X-ray photoelectron spectroscopy (XPS, AXISULTRA-DLD, Thermo Fisher, Tokyo, Japan) was employed to investigate the chemical bonding states. The O_1s XPS spectra were analyzed by using an X-ray source of Al Kα monochromater and calibrated by the C_1s line of carbon. Shirley’s method was used to subtract the background, and a nonlinear and least squared routine with a mixed Gauss–Lorentz function was employed to fit the peaks.

2.3.5. Physical Properties Test of Cast Stone

The cast stone samples obtained by casting were processed and cut. The flexural strength, water absorption rate, and bulk density were tested according to the building materials standard (GB/T 18601-2009 “Natural Granite Building Panels” [26]). The flexural test samples were 3 mm × 4 mm × 10 mm in size and tested on an electronic universal testing machine (TZS-6000, Xian, China). The flexural performance was evaluated using a four-point bending method with a crosshead speed of 0.5 mm min⁻¹. The water absorption rate and bulk density of the samples were measured using a ceramic water absorption vacuum device (CXK-A, Guangdong, China) and a precision ceramic bulk.

3. Results and Discussion

3.1. Mechanical Properties of Cast Stone

The cast stone samples of silicomanganese slag, prepared in the laboratory according to three different formulations with dimensions of 40 mm × 40 mm × 10 mm, are shown in Figure 4. Unlike the other two samples, S1 exhibited a glassy luster on its cross-section, which was subsequently confirmed to consist almost entirely of an amorphous phase. As shown in

Table 3. Mechanical properties of samples and national standard requirements.

Sample/National Standard	Flexural Strength/MPa	Bulk Density/(g/cm−3)	Water Absorption/%
S1	—	3.03	0.08
S2	37.50	3.11	0.32
S3	35.70	3.09	0.22
Natural granite building slabs (GB/T 18601-2009) [26]	≥8.3	≥2.56	≤0.4
Ceramic tiles (GB/T 4100-2015) [27]	≥35	—	≤0.5

, the flexural strength, bulk density, and water absorption of the modified cast stone samples S2 and S3, prepared from modified silicomanganese slag, were 37.50 MPa and 35.70 MPa, 3.11 g cm⁻³ and 3.09 g cm⁻³, and 0.32% and 0.22%, respectively. These values meet the requirements of GB/T 18601-2009 “Natural granite building Panel” [26] and GB/T 4100-2015 “Ceramic tiles” [27]. Notably, following the formulation of S2, silicomanganese slag cast stone raw stone was successfully prepared on a factory scale, as illustrated in Figure 4.

In this study, the method of utilizing sensible heat to directly melt the modifier was primarily aimed at providing a novel approach to address the bulk utilization of the silicomanganese slag. By achieving compliance with building material standards, the sensible heat in silicomanganese slag was effectively harnessed through simple modification to produce cast stone with low production costs. This work establishes a theoretical foundation for industrial-scale implementation.

3.2. Analysis of Crystallization Performance of Silicomanganese Slag

3.2.1. Crystallization Temperature

The crystallization behavior of the silicomanganese slags was investigated using non-isothermal DSC method. As shown in Figure 5, the DSC curves of the modified silicomanganese slag samples S2 and S3 exhibited pronounced exothermic peaks across the three different cooling rates (10, 20, and 30 °C min⁻¹) (see

Table 4. The exothermic peak (T_p) of samples at different cooling rates.

Cooling Rate β (°C·min−1)	Tp (°C)
	S1	S2	S3
10	–	1081	1114
20	–	1073	1093
30	–	1045	1050

Note: The exothermic peak of raw silicomanganese slag in Figure 5a is hard to be read out, so it is recorded as “–”.

), which is indicative of their strong propensity for crystallization. In contrast, the DSC curve of the raw slag sample S1 showed negligible exothermic peaks under the same cooling rate conditions, suggesting that the raw silicomanganese slag exhibits poor crystallization behavior. This observation underscores the beneficial effects of slag modification in promoting crystal formation, which is a key rationale for the modification efforts in this study.

The influence of cooling rate on the shape and position of the exothermic peaks is also evident. As shown in Figure 5b,c, lower cooling rates resulted in sharper peak profiles, higher peak intensities, and elevated peaks tip temperatures. Conversely, higher cooling rates lead to blunter and lower exothermic peaks, and in some cases, the exothermic peaks became difficult to discern. This phenomenon can be attributed to the high silicon content in the raw slag, which hinders the crystallization. While modifiers are employed to promote crystallization, higher cooling rates often result in insufficient crystallization, thereby reducing the crystallization exothermicity. Additionally, since the nucleation temperature of crystals is typically lower than the growth temperature [28], higher cooling rates require a greater driving force to effectively promote nucleation and crystal growth [29,30,31]. As the cooling process continues and the temperature gradually decreases, the ion migration rate also diminishes, posing a significant challenge to crystal growth.

3.2.2. Crystallization Activation Energy

The aforementioned analysis demonstrates that incorporating various modifiers significantly influences the crystallization process of silicomanganese slag, potentially inducing structural modifications and altering the energy barriers required for atomic rearrangement during crystallization. These factors ultimately determine the differential crystallizability observed among the samples. Such crystallization processes can be effectively analyzed through crystallization kinetic methodologies [31,32,33]. Numerous models have been developed to elucidate the crystallization activation energy and crystallization index of metallurgical slag and polymers under non-isothermal conditions. In this study, the analysis was conducted by Kissinger and Augis–Bennett equations [34]. The Kissinger equation is given by Equation (1):

$$\ln \left({T_{\mathrm{p}}}^2/\beta \right)=E_{\mathrm{a}}/\mathrm{R}{T}_{\mathrm{p}}+\mathrm{C}$$

(1)

where E_a (kJ mol⁻¹) is the crystallization activation energy, β (°C min⁻¹) is the cooling rate, T_p (°C) is the maximum crystallization peak temperature value, R (8.314 J/(mol K)) is the gas constant, and C is a constant.

According to Equation (1), a linear relationship between ln (T_p²/β) and 1/T_p was established, as illustrated in Figure 6, where discrete data points were fitted with a linear regression curve. The slope of the fitted line corresponds to the crystallization activation energy (E_a), while the Pearson correlation coefficient (r) indicates the goodness of fit, both of which are explicitly presented in Figure 6. Quantitative analysis revealed that the crystallization activation energies for samples S2 and S3 were determined to be 407.24 kJ mol⁻¹ and 237.42 kJ mol⁻¹, respectively. This is attributed to the increased content of nucleating agents in the modified silicomanganese slag, which promotes non-uniform nucleation of the silicomanganese slag, and the melting point of the spinel generated by S3 being higher than that of S2, resulting in easier crystallization of S3.

3.2.3. Crystalline Phases

To investigate the crystallization behavior of silicomanganese slag under two modification methods, the samples were first heated to 1500 °C to achieve complete melting, followed by controlled cooling to 1300 °C, 1100 °C, and 1100 °C with a 30 min holding period at each temperature before quenching. The corresponding XRD results are shown in Figure 7, demonstrating excellent agreement with the non-isothermal DSC findings and providing comprehensive insights into the crystallization characteristics of the studied slags. Both analytical techniques consistently confirm that S1 exhibits limited crystallization tendencies. The XRD analysis further reveals that S3 and S2 display nearly identical diffraction peak positions at 1000 °C, indicating the formation of identical crystalline phases in both samples. Detailed examination of the diffraction peak intensities and shifts demonstrates that the primary crystalline phases in both S2 and S3 are augite, accompanied by spinel-group minerals including galaxite ((Mn²⁺, Mg)(Al, Cr³⁺, Fe³⁺)₂O₄) and normal spinel (Mg(Cr³⁺, Al)₂O₄) at 1300 °C and 1100 °C, which subsequently disappear at 1000 °C. Additionally, akermanite-gehlenite is observed to coexist with augite at 1000 °C.

The observed variations in the formation of galaxite ((Mn²⁺, Mg)(Al, Cr³⁺, Fe³⁺)₂O₄) and normal spinel Mg(Cr³⁺, Al)₂O₄, both belonging to the spinel group of minerals, can be attributed to differences in the slag’s chemical composition. The formation of these spinel phases follows the typical chemical reaction: Mn²⁺+Mg²⁺+Fe²⁺+2Cr³⁺+4O²⁻ → (Mn²⁺, Fe²⁺, Mg²⁺)(Al³⁺, Cr³⁺, Fe³⁺)₂O₄. The relatively higher concentrations of Mn²⁺ and Fe^2+/3+ in S2, combined with the elevated Mg²⁺ content in S3, create thermodynamically favorable conditions for the crystallization of galaxite and normal spinel. Through strategic modification of the silicomanganese slag’s chemical composition by introducing Cr³⁺, Fe^2+/3+, and Mg²⁺ ions, these ions effectively combine with Al³⁺ and Mn²⁺ ions present in the slag to form galaxite and normal spinel. Importantly, both galaxite and spinel function as efficient nucleating agents, promoting the generation of augite and thereby enhancing the overall crystalline properties of the silicomanganese slag, which aligns with previous research findings [35,36].

In the context of producing casting stone from silicomanganese slag, the presence of finely dispersed spinel particles does not significantly compromise the fluidity of the molten slag due to their minimal size, while simultaneously promoting the formation of augite. Consequently, the enhanced crystallization capability of silicomanganese slag significantly improves its suitability for conversion into casting stone. These findings highlight the substantial benefits of optimizing the crystallization properties of silicomanganese slag for its downstream industrial applications.

3.3. Analysis of Fluidity Performance of Silicomanganese Slag

3.3.1. Viscosity of Silicomanganese Slag

Figure 8 shows the measured viscosity of different silicon manganese slags at different temperatures. The results reveal that the viscosity of silicomanganese slag at the same temperature follows the order S1 < S2 < S3, attributable to the gradual decrease in R value (mass%CaO/mass%SiO₂) and the relative content of Ca²⁺, K⁺, Mg²⁺, and Fe²⁺ acting as fluxing agents, leading to elevated viscosity and increased complexity of the aluminosilicate structure. This trend aligns with findings reported by several researchers [37,38,39,40]. The viscosity–temperature profiles of modified silicomanganese slags display relatively gentle slopes, with all curves approximating hyperbolic shapes, which is characteristic of long slag (acidic slag) behavior. Notably, as the temperature decreases to the pouring temperature of the slag (1340~1400 °C), the viscosity–temperature curves of the silicomanganese slag exhibit minimal indication of turning points caused by the generation of spinel-group minerals. The absence of crystal phase generation in S1 can be attributed to its high viscosity and the insufficient content of existing nucleating agents to promote crystal precipitation. Despite the enhancement in viscosity of S2 and S3, their crystallization performance remains superior to that of silicon manganese slag S1. This suggests that the modifier plays a more prominent role in decreasing the activation energy required for crystallization, outweighing the inhibitory effects of increased viscosity on crystallization.

The production of cast stone involves transforming slag into a solid geometric form, which requires the material to maintain appropriate fluidity within a specific temperature range to ensure optimal formability [41]. Through linear regression analysis of the viscosity curves presented in Figure 8 within the temperature range of 1340–1400 °C, we observed a gradual increase in the slopes for samples S1, S2, and S3, with corresponding values of −0.00914, −0.00899, and −0.00887, respectively. These findings demonstrate that the viscosity of modified silicomanganese slag exhibits a more gradual variation with decreasing temperature compared to its unmodified counterpart. The modification of slag to achieve a long slag characteristic proves particularly advantageous for cast stone production, as it not only mitigates abrupt viscosity changes induced by crystallization but also provides stress-softening capabilities. This dual benefit effectively inhibits defect formation during the cooling process, thereby enhancing the overall product quality. Specifically, modified silicomanganese slags S2 and S3 exhibit superior crystallization properties while maintaining excellent fluidity throughout the casting process. The slow and fine crystallization process inherent in these modified slags effectively prevents viscosity mutations associated with crystallization, which significantly contributes to the optimization of the casting process.

3.3.2. Activation Energy for Viscous Flow of Silicomanganese Slag

The activation energy for viscous flow of molten slag is an important parameter indicator to describe the viscous behavior of slag and is an energy barrier that the cohesive flow units in slag must overcome when moving between different equilibrium states [42]. It is usually expressed by the Arrhenius formula (Equation (2)).

$$\eta =A\exp \left(E_{\eta }/RT\right)$$

(2)

where η is the viscosity, Pa s; A is the pre-exponential factor; E_η is the activation energy for viscous flow, J mol⁻¹; R is the universal gas constant, J mol⁻¹ K⁻¹; and T is the absolute temperature, K.

Taking the logarithm of Equation (2) gives the following expression:

$$\ln \eta =\ln A+E_{\eta }/RT$$

(3)

Figure 9 presents the fitted relationship between the lnη and 1/T. Based on the experimental data, the activation energy for viscous flow E_η can be obtained from the slope of the fitted straight line. As can be seen in Figure 9, lnη has a good linear relationship with 1/T, and the linear dependence coefficient of each fitted line is higher than 0.99. The calculated activation energy for viscous flow is shown in

Table 5. Activation energy for viscous flow of different modified silicomanganese slags.

Slag No.	S1	S2	S3
Activation energy for viscous flow (kJ mol−1)	30.38 ± 8.34	43.70 ± 5.14	74.93 ± 8.07

. The activation energy for viscous flows before and after the modification in Table 5 were 30.38 ± 8.34 kJ mol⁻¹ (S1), 43.70 ± 5.14 kJ mol⁻¹ (S2), and 74.93 ± 8.07 kJ mol⁻¹ (S3), respectively. This indicates an increase in the energy barrier for viscous flow and the formation of some more complex structural units in the slag, which results in a higher viscosity of slag melts. The activation energy for viscous flow values and viscosity values showed the same trend, indicating that the viscous flow resistance of silicomanganese slag increased after modification.

3.4. The Relationship Between Performance and Structure of Cast Stone

3.4.1. FTIR Spectra Analysis

Figure 10 presents the FTIR spectra of the raw and modified silicomanganese slag samples within the 1400–400 cm⁻¹ shift range, which can be divided into two distinct regions: the high-frequency region (1200–800 cm⁻¹) and the low-frequency region (800–600 cm⁻¹). The characteristic peaks in these shift ranges have been attributed based on related research by other authors, as outlined in

Table 6. Assignments of FTIR bands in spectra of silicomanganese slags.

Wavenumber/cm−1	Assignment	Refs.
600–800	[AlO4]-tetrahedral stretching vibration	[44,45,46]
800–1200	[SiO4]-tetrahedral stretching vibration	[44,47,48]

. The presence of [SiO₄]-tetrahedra signifies the network structure of the glass, while the widened absorption band in the 1200–800 cm⁻¹ range is associated with the stretching vibration of the [SiO₄]-tetrahedral Si–O bonds influenced by varying quantities of bridging oxygen Qⁿ (n = 0, 1, 2, 3, and 4). Additionally, Al³⁺ ions function as glass network intermediates and occupy the [AlO₄]-tetrahedral structure formed by non-bridging oxygen [43], with absorption bands in the 800–600 cm⁻¹ range attributed to [AlO₄]-tetrahedral stretching vibration.

This analysis places greater emphasis on the stretching vibration of the [SiO₄]-tetrahedron due to the relatively higher SiO₂ content in the samples, resulting in a correspondingly stronger absorption band associated with the [SiO₄]-tetrahedron. Notably, the absorption bands corresponding to the [SiO₄]-tetrahedral of samples S2 and S3 are observed to shift toward the high-frequency region compared to S1, indicating an increase in the number of [SiO₄]-tetrahedra with more bridging oxygen. This suggests an increase in the polymerization of the modified silicomanganese slag samples S2 and S3, contributing to elevated mobility temperature and viscosity of the modified slag.

3.4.2. XPS Spectra Analysis

In order to delve deeper into the mechanism of viscosity effects on the silicomanganese slag structure, the variation in the molar fraction of oxygen atoms (bridging oxygen (O⁰), non-bridging oxygen (O⁻), and free oxygen (O²⁻)) was analyzed for different samples. Previous research by various scholars [49,50,51] has indicated that the peaks located at binding energies of approximately 534 eV, 532 eV, and 531 eV in O_1s XPS spectra correspond to O⁰, O⁻, and O²⁻, respectively. Figure 11a–c depict the results of the deconvolution spectra of samples S1, S2, and S3 based on XPS PEAK4.1, respectively. The relative area fraction serves as a stable representation of the content of individual oxygen constituents, enabling the analysis of structural evolution using this method. The relative contents of each structural unit obtained from the sample deconvolution spectra are presented in Figure 12. It is evident that the fraction of O⁰ increases and O⁻ decreases after modification, with the change being more pronounced for S3 than for S2. This indicates an increase in the degree of slag polymerization, consistent with the FTIR analysis. The degree of polymerization and viscosity of the slag are closely linked, and both FTIR and XPS results confirm the increased viscosity of the modified samples S2 and S3.

The relationship correlating O²⁻, O⁻, and O⁰ at high temperatures can be described by O⁰ + O²⁻ ⇋ 2O⁻, where O²⁻ is provided by basic oxides such as CaO, MgO, MnO, K₂O, and BaO in this system. However, the R value decreased after modification, resulting in lower content of O²⁻, leading to an increase in O⁰ and complicating the network, thereby increasing the viscosity. Nevertheless, the increase in viscosity after modification is not conducive to the nucleation and growth of crystals, which contradicts the DSC results. On one hand, this contradiction is attributed to the presence of nucleating agents in the modifier, which facilitates the reduction of crystallization activation energy, promoting crystal nucleation and the generation of fine spinel phase. On the other hand, the lowered R value in the slag after modification increases O⁰, thus augmenting the degree of polymerization and viscosity of the slag, which is not conducive to the precipitation of crystals.

4. Summary

The experiments involved modifying silicomanganese slag by adding 10 mass% cold chromite and serpentine, and 20 mass% molten ferrochrome slag, resulting in samples named S2 and S3, respectively. S1 was unmodified silicomanganese slag.

The main mineral phases of the modified silicomanganese slag were identified as augite and spinel-group minerals, including galaxite (Mn²⁺, Mg)(Al, Cr³⁺, Fe³⁺)₂O₄) and normal spinel (Mg(Cr³⁺, Al)₂O₄)). The crystallization activation energies were 407.24 kJ mol⁻¹ for S2 and 237.42 kJ mol⁻¹ for S3. The precipitation of spinel-group minerals was observed in the quenched slag at 1300 °C and 1100 °C. These minerals played a dual role: acting as a nucleation agent to dynamically promote the formation of augite, and increasing the proportion of SiO₂ and polymerization of [SiO₄] structural units in the residual liquid phase, further enhancing the generation of augite in terms of composition. The gradual precipitation of crystals effectively mitigated sudden viscosity fluctuations caused by crystallization. As a result, the modified slag exhibited both superior crystallization ability and favorable fluidity.

The results indicated that the cast stone made from raw silicomanganese slag remained primarily amorphous and fractured after heat treatment. In contrast, the crystallization ability of the two modified slags was significantly enhanced, resulting in cast stone with overall good performance. Their physical properties, including flexural strength, bulk density, and water absorption, meet the requirements of the GB/T18601-2009 standard for “Natural Granite Building Slab” [26]. Specifically, the flexural strengths were 37.50 MPa and 35.70 MPa, the bulk densities were 3.11 g·cm⁻³ and 3.09 g·cm⁻³, and the water absorption rates were 0.32% and 0.22%, respectively.

This study demonstrated the feasibility of a simple and low-cost modification method using either “molten silicomanganese slag + 10 mass% cold modifiers” or “molten silicomanganese slag + 20 mass% molten ferrochrome slag”, relying on the self-induced heat of the molten silicomanganese slag.