2022 Volume 62 Issue 5 Pages 941-947
In developed countries, nearly half of steelmaking slag is reused in road construction. Determining free lime content in steelmaking slag is important because the hydration of free lime can result in the road fracturing. Recently, we have developed a simple method to rapidly determine the area fraction of free lime via cathodoluminescence (CL) imaging of steelmaking slag based on orange illuminated areas corresponding to free lime. However, the areas of free lime that precipitated along grain boundaries of wüstite were overestimated in CL images because the thickness of the precipitated free lime was not evaluated properly. In this study, we investigate the grain-boundary microstructures of wüstite and propose a method to correct the overestimated areas of the free lime. Scanning transmission electron microscope observation revealed that the average thickness of free lime precipitated in the grain boundaries was 250 nm, which was approximately 20 times smaller than that evaluated via the CL image. Assuming that the thickness of the free lime precipitated in the grain boundaries was 250 nm as a whole, the area fractions decreased by 60%–80%, compared with those without the thickness correction. Considering the CL sampling depth of a few microns, the proposed method can determine the volume fraction of free lime in steelmaking slag more precisely, hereby aiding in the prevention of road expansion in the reuse of steelmaking slag.
Steelmaking slag is a byproduct that is generated during the conversion of carbon-rich molten iron or scrap into steel conducted in basic oxygen furnaces (BOFs) or electric arc furnaces, respectively, as well as during the refining of steel in ladle furnaces. Steelmaking slag accounts for approximately 12–20 mass% of total steel production,1,2) indicating that a huge amount of steelmaking slag is produced worldwide. To accommodate the current environmental requirements for minimizing waste disposal and achieving an environmentally friendly society, iron and steel manufacturing industries have actively worked toward the reuse of steelmaking slag. A high reuse rate of steelmaking slag is achieved in developed countries, such as Japan (98.4%), Europe (87%), and the United States (85%).2,3,4) In these countries, steelmaking slag has been mostly reused in road construction, e.g., to build road bases, subbase courses, and pavement surfaces (32% in Japan, 43% in Europe, and 60% in the United States).2,3,4,5,6,7,8,9) Steelmaking slag usually contains 2–8 mass% of free lime,6) which remains in the form of CaO without reacting with other compounds present in the slag. Free lime doubles its volume upon reacting with the moisture in the air, and the reaction rate is slow,2,7,10) thereby causing road fracturing. Therefore, an upper limit is charged on the amount of free lime in steelmaking slag, or slag conditioning is required for road construction application.2,11) In slag conditioning, the hydration reaction of free lime is accelerated by exposing the slag to air for several months, spraying it with hot water, or steaming it for a week.6,11) However, if the slag conditioning is insufficient, the residual free lime can induce road expansion and fracturing.2,8,12) Therefore, determining free lime content is paramount to safely use steelmaking slag for road construction.
Several methods to determine the free lime content in steelmaking slag have been presented, such as ethylene glycol extraction followed by the determination of the calcium content in ethylene glycol via titration, atomic absorption spectroscopy, or inductively coupled plasma-atomic emission spectrometry (ICP-AES),6,13,14,15) X-ray diffraction,16,17,18,19,20,21) and infrared spectrometry.22) Among these methods, the method combined ethylene glycol extraction and ICP–AES is the most commonly used. This method is a complicated and time-consuming procedure that requires skilled operations, e.g., controlling the particle size of the steelmaking slag and monitoring the temperature and extraction time.6,15) Therefore, a simple method to rapidly determine the free lime content in steelmaking slag is required.
Recently, we have developed a method for determining the free lime content in industrial steelmaking slag via cathodoluminescence (CL) imaging, which provided a projection map of compound phases based on the phenomenon of light emission by electron bombardment.23) This method was based on measuring orange illuminated areas corresponding to free lime and was simpler and quicker than the one involving ethylene glycol extraction. The orange luminescence of free lime originated from the manganese (II) ions (Mn2+) dissolved into free lime. However, free lime with small sizes was overestimated in the CL images, thereby overestimating the free lime contents.23) Free lime in steelmaking slag is classified into five groups: (I) unassimilated CaO (unmelted CaO during the steelmaking process.), (II) CaO crystallized with fine precipitates like wüstite, (III) CaO produced by the decomposition of 3CaO·SiO2 into 2CaO·SiO2, (IV) primary CaO, which is normally observed in steelmaking slag with the basicity (CaO/SiO2 ratio) >4.0, and (V) CaO precipitated along the grain boundaries of wüstite or 2CaO·Fe2O3.24,25) Areas of free lime in (I)–(IV) can be evaluated from CL images because the grain sizes of such free limes are large enough to be identified via CL images (greater than a few micrometers). Even though the Mn content of the unassimilated CaO is lower than that of other groups of free lime, luminescence originating from Mn2+ in the unassimilated CaO can be detected because CaO including only parts-par-million (ppm) levels of Mn2+ exhibit the CL peak owing to Mn2+.26) We could detect orange luminescence originating from CaO crystallized with fine precipitates (areas A–G in Fig. 1(a)) and CaO produced by the decomposition of 3CaO·SiO2 to 2CaO·SiO2 (areas H and I in Fig. 1(b)) in an industrial steelmaking slag sample by acquiring CL images. These illuminated areas agreed well with the areas of free lime observed in the scanning electron microscopy (SEM) image and elemental mappings (Figs. 1(b)–1(e)). Areas emitting orange luminescence with identical CL spectra as free lime were detected in the grain boundaries of wüstite in the CL images (areas i–v in Fig. 1(a)). However, we could not identify the free lime in these areas through SEM–energy dispersive X-ray (EDX) analysis because high intensities of characteristic X-rays resulting from calcium (Ca Kα line) were not detected (Fig. 1(c)), which indicated that the thickness of free lime precipitated along the grain boundaries of wüstite was too small to detect via SEM–EDX analysis. Despite the results of SEM–EDX analysis, the thickness of the free lime could be estimated as several micrometers from the CL images. This difference in the thickness of free lime precipitated along the grain boundaries of wüstite overestimated the area fractions of free lime via CL imaging.
In this study, we investigate the grain-boundary of wüstite and propose a method to correct the area fractions of free lime precipitated in the grain boundaries of wüstite. For this purpose, we examined the grain-boundary microstructures of wüstite that emitted orange luminescence via scanning transmission electron microscopy (STEM). The results of our research facilitate the safe use of steelmaking slag in road construction.
BOF slag by-produced by Nippon Steel Corp. was analyzed in this study. None of the slag samples underwent slag conditioning. The free lime content of these industrial slag samples was 3.33 mass%–determined by the ethylene glycol extraction method using ICP–AES.23) The basicity was 3.41. The concentrations of components (e.g., CaO, SiO2, Al2O3, MgO, MnO, P2O5, S, and total Fe) measured using ICP–AES were similar to those reported for typical BOF slags.6,27,28) The XRD pattern was shown in Fig. 2, and the detailed properties described in our previous report.23) The sample surface was polished using 600-, 1200-, and 2400-grit abrasive papers and finished using water-free 1-μm diamond slurry (Aka-Poly WF 1 μm, Akasel A/S, Roskilde, Denmark). Then, the samples were placed for 5 min in a box furnace pre-heated at 1000°C in air, and successively extracted the samples and blown with argon gas for quenching to room temperature to enhance the CL intensity of free lime.23,29)
XRD pattern of the slag sample.23)
The samples were observed and characterized using a STEM (EM-002B, Topcon Corp., Tokyo, Japan) equipped with a lithium-drifted silicon (Si (Li)) EDX detector (NORAN System 7, Thermo Fisher Scientific Inc., Massachusetts, USA), a SEM (TM3030 Plus, Hitachi High-Technologies Co., Tokyo, Japan) equipped with a silicon drifted EDX detector (Quantax70, Bruker Corp., Billerica, Massachusetts, USA), a field emission electron probe microanalyzer (FE-EPMA) (JXA-8530F, JEOL Ltd., Tokyo, Japan) equipped with a wavelength dispersive X-ray spectrometer. For the STEM observation, we cut the areas of the slag samples marked by the broken lines of areas 1 and 2 in Fig. 1(a) using a focused ion beam (FIB) instrument (Helios NanoLab 600i, FEI Co., Hillsboro, Oregon, USA). The STEM microscope’s acceleration voltage was set to 200 kV. The CL images were acquired using a customized SEM–CL system, which is described in our previous reports.23,26,30,31,32,33,34,35,36,37,38,39,40,41,42,43) In brief, the CL images were captured through a quartz viewport attached to a SEM instrument (Mighty-8DXL, TECHNEX, Tokyo, Japan) using a digital mirrorless camera (α7RII, Sony Corp., Tokyo, Japan) equipped with a zoom lens (LZH-10A-05T, Seimitu Wave Inc., Kyoto, Japan). The detectable wavelength range of this camera was 420–680 nm. The acceleration voltage of the SEM instrument was set to 17 kV.
Orange luminescence with identical CL spectrum as free lime was detected in several areas along the grain boundary between wüstite and 2CaO·SiO2 (Fig. 1(a)). We investigated the microstructures of areas including the grain boundary between wüstite and 2CaO·SiO2 with (area 1 in Fig. 1(a)) and without orange luminescence (area 2 in Fig. 1(a)) via STEM to understand the difference in the luminescence behavior of these areas. Figure 3(a) shows the STEM cross-section image of area 1, and EDX elemental mappings of Ca, Si, and Fe are shown in Figs. 3(b), 3(c), and 3(d), respectively. Wüstite corresponds to an area where only Fe was detected (left side of Fig. 3(a)), whereas 2CaO·SiO2 corresponds to an area where Ca and Si were detected (right side of Fig. 3(a)). An area where the intensity of Ca Kα line was higher than that of 2CaO·SiO2 was detected along the grain boundary between wüstite and 2CaO·SiO2 (Fig. 3(b)). Neither Si nor Fe was not detected in this area (Figs. 3(c) and 3(d)). This result indicates that free lime was precipitated along the grain boundary in area 1. By contrast, for area 2 (emitting no luminescence), an area where the intensity of Ca Kα line was higher than that of 2CaO·SiO2 was not observed along the grain boundary between wüstite (left side in Fig. 4(a)) and 2CaO·SiO2 (right side in Fig. 4(a)). However, calcium was detected in a few areas in wüstite (Fig. 4(b)). This calcium in the areas originated from 2CaO∙SiO2 that was adhered to the areas during the sputtering of samples with the gallium ion beam in the FIB instrument to make the sample thinner for the STEM observation. These results indicate that free lime was not precipitated along the grain boundary in area 2. Therefore, the STEM analysis revealed that free lime precipitated along the grain boundary between wüstite and 2CaO·SiO2 in area 1 was responsible for the orange luminescence in the CL image (Fig. 1(a)).
(a) STEM (bright field) image and EDX elemental mappings of (b) Ca, (c) Si, and (d) Fe for the cross-section of area 1 in Fig. 1(a).
(a) STEM (bright field) image and EDX elemental mappings of (b) Ca, (c) Si, and (d) Fe for the cross-section of area 2 in Fig. 1(a).
CL images could detect free lime that was precipitated along the grain boundary between wüstite and 2CaO·SiO2, whereas SEM–EDX analysis could not (Fig. 1(c)). Even an EPMA line scan profile at higher magnification could not exactly detect the precipitated free lime, as shown in Fig. 5; increase of the Ca Kα line intensity near the grain boundary between wüstite and 2CaO·SiO2 with orange luminescence was not detected. These results indicate that acquiring CL images is a method to easily and rapidly identify free lime precipitated along a grain boundary of wüstite, which is difficult to achieve via SEM–EDX and EPMA analysis.
EPMA line scan profiles of Ca and Fe Kα lines along the arrow of (a) α, (b) β, (c) γ, and (d) δ in Fig. 1(b).
STEM results confirmed that orange luminescence in the grain boundary between wüstite and 2CaO·SiO2 originated from free lime. The thickness of the precipitated free lime in the grain boundary was approximately 250 nm (Fig. 3). However, the thickness of the free lime in the grain boundary was approximately 5.5 μm from CL image (Fig. 1(a)). This huge difference in the free lime thicknesses caused an overestimation of free lime content, which was determined from fractions of areas with orange luminescence in CL images. Thus, we corrected the overestimated area fractions of free lime under the assumption that the thickness of all the free lime precipitating at the grain boundary between wüstite and 2CaO·SiO2 was 250 nm. The area fraction of free lime in Fig. 1(a) was corrected to 31.1%, which was 0.83 times smaller than that before the correction (Table 1). Area fractions of free lime in other areas of the slag sample including free lime precipitated along the grain boundary were determined from their CL images (Figs. 6(a) and 6(c)) based on the correction, and the results are summarized in Table 1. Free lime precipitated along the grain boundary in the CL images was identified from their SEM images (Figs. 6(b) and 6(d)). These area fractions of free lime also resulted in 0.60–0.65 times decrease due to the correction. Our previous study indicated that free lime precipitated along the grain boundary was observed in slag samples with high free lime contents.23) In addition, it is well known that industrial steelmaking slag has a large variation in composition among lumps. Thus, these corrected area fractions of free lime could be higher than the average area fraction. Areas with a low area fraction of free lime were often observed in other slag samples of different lumps (Fig. 7). The average area fraction of free lime was evaluated as 7.2 ± 3.3%, whereas that without the correction was 9.8 ± 2.7%.23) Five slag samples of different lumps were used for measuring the area fractions, and the CL images were captured for at least five areas of each sample (Table 2). We considered that the area fractions of free lime were unaffected by the thickness of free lime and thus were almost identical to volume fractions of free lime in the slag sample because the detection depth of the CL analysis was a few microns at the accelerating voltage of 17 kV.44,45)
(a) and (c) CL images and (b) and (d) the corresponding SEM image of other specimens in the batch of the slag sample. The exposure times for the CL images were 5 s. (Online version in color.)
Typical CL images of the slag sample. (Online version in color.)
| Area No. | Sample No. | ||||
|---|---|---|---|---|---|
| #1 | #2 | #3 | #4 | #5 | |
| #I | 31.1 | 9.2 | 0.3 | 0.2 | 17.8 |
| #II | 11.9 | 1.0 | 0 | 0 | 8.6 |
| #III | 17.3 | 2.0 | 0 | 0.1 | 16.3 |
| #IV | 20.7 | 1.4 | 0.2 | 0.1 | 3.6 |
| #V | 34.0 | 8.4 | 0.1 | 0.1 | 0.3 |
| #VI | – | 4.3 | – | – | 6.5 |
| #VII | – | – | – | – | – |
| Average | 23.0 | 4.1 | 0.1 | 0.1 | 8.9 |
| Standard deviation | 8.4 | 3.1 | 0.1 | 0.1 | 6.3 |
– means not measured.
A cooling rate for steelmaking slag varies with batches of industrial steelmaking slag. Even for the same batch of industrial steelmaking slag, the cooling rate can vary with its location in furnaces. Because steelmaking slag in the upper part of the furnaces is exposed to the air, its cooling rate is larger than that for steelmaking slag in the lower part. This difference in the cooling rate would cause the difference in the thickness of the precipitated free lime along the grain boundary between wüstite and 2CaO·SiO2; the thickness of the precipitated free lime is smaller for steelmaking slag in the upper part of the furnaces, compared with the thickness of steelmaking slag in the lower part. To investigate the effect of the thickness of the precipitated free lime along the grain boundary on the area fraction of free lime in steelmaking slag, we calculated area fractions of free lime in Fig. 1(a), Figs. 6(a) and 6(b) under the assumption that the thicknesses of all the precipitated free lime along the grain boundary between wüstite and 2CaO·SiO2 were 100 and 500 nm, as shown in Table 3. The differences in the area fractions were within 0.6%, which is more than 10 times lower than the differences in the area fractions between with and without the correction. This indicates that the present thickness correction method for the precipitated free lime along the grain boundary is not significantly affected by the cooling rate of steelmaking slag.
Next, we converted the obtained area fraction of free lime in Fig. 6(a) to a mass concentration based on the results of XRD (Fig. 2) and EDX elemental mappings (Fig. S1), and compared the mass concentration with the free lime content measured by ICP–AES (3.33 mass%). The results of XRD and EDX elemental mappings suggested the presence of 2CaO·SiO2, wüstite, and Ca2(Al,Fe)2O5 phases besides free lime phase. Area fractions of 2CaO·SiO2, wüstite, and Ca2(Al,Fe)2O5 phases were roughly estimated form the EDX elemental mappings (Fig. S1). The area fractions and the area fraction of free lime were then converted to mass concentrations using a reported density of each mineral phase46,47) under the assumption that the area fraction of each mineral phase was almost identical to a volume fraction, as shown in Table 4. The mass concentration of free lime acquired using the thickness correction was closer to that measured by ICP–AES than that without the thickness correction. This indicates that we can roughly determine the volume fraction of free lime in the slag sample by correcting the areas fraction of free lime precipitated along the grain boundary between wüstite and 2CaO·SiO2. However, the corrected mass concentration of free lime (11 mass%) remained higher than that measured by ICP–AES (3.33 mass%). This may be because the densities of 2CaO·SiO2, wüstite, and Ca2(Al,Fe)2O5 phases in Fig. 6(b) were higher than the reported values owing to the dissolution of iron and manganese into these mineral phases. The presence of miner phases that was not detected using XRD may also be responsible for the higher mass concentration of free lime. Nevertheless, our method is promising in that we can determine the volume fraction of free lime (not the mass concentration) because the volumetric expansion of free lime is a major issue for the reuse of steelmaking slag in road construction. To the best of our knowledge, the method proposed in this study is the first to estimate the volume fraction of free lime in steelmaking slag.
| Mineral phase | Area fraction (%) | Density (g cm–3) | Mass concentration (mass%) |
|---|---|---|---|
| 2CaO·SiO2 | 53.6 | 3.3146) | 49.2 (46.5) |
| Wüstite | 22.4 | 5.70 | 35.5 (33.5) |
| Ca2(Al,Fe)2O5 | 4.1 | 3.6847) | 4.2 (4.0) |
| Free lime | 11.9* (18.2**) | 3.35 | 11.1 (16.0) |
We investigated the origin of orange luminescence in the grain boundary between wüstite and 2CaO·SiO2 in industrial steelmaking slag after quenching from 1000°C to room temperature by observing the grain-boundary microstructures via STEM. Free lime was detected in the grain boundary emitting orange luminescence, whereas none was detected in the grain boundary emitting no luminescence. The thickness of the free lime precipitated in the grain boundary was approximately 250 nm, which was approximately 20 times smaller than that estimated from CL images. Based on the measured free lime thickness, we proposed a method to correct area fractions of free lime precipitated in grain boundaries, which contributes to a precise estimation of the amounts of free lime in steelmaking slag. The area fractions were 60%–80% smaller than those before the correction. The average exposure time for CL images with the areas of 0.5 × 0.1 mm was 5 s. Therefore, the proposed method is a promising technique to rapidly determine volume fractions of free lime in industrial steelmaking slag, thereby preventing road fracturing.
SEM image and EDX elemental mapping of Mg, Al, Si, Ca, Mn, and Fe for the area in Fig. 6(b).
This material is available on the Journal website at https://doi.org/10.2355/isijinternational.ISIJINT-2021-465.
This study was supported by the 30th ISIJ Research Promotion Grant. We would like to thank Nippon Steel Corp. for the supplies of steelmaking slag sample and for the determination of its free lime content. The authors thank Mr. Shun Itoh and Ms. Kaori Satoh of Tohoku University for their help in performing STEM observation and FIB fabrication, respectively.
