**3. Results**

After cooling the tapped slag, the discarded slag showed that the produced metal settled properly (see Figure 7B) at the bottom of the ladle and could, as intended, be easily separated from the slag. The slag, as shown in Figure 7, also showed the expected structures. The outermost slag shell was amorphous and glass-like and looked like a black–green basalt glass. The slag of zone 2 showed no amorphous phases. Here, small metallic inclusions were very easily visible. In the innermost zone 3, after 24 h of cooling (approximately 80 ◦C), the slag mostly showed the same structure to that in the intermediate zone, but with a more pronounced crystallinity.

**Figure 7.** (**A**) Molten slag phase inside the TBRC during the melting process; (**B**) cooled slag and metal block [5].

After cooling and crushing, the slag was subsequently analyzed. Therefore, several samples from di fferent positions inside the ladle were taken, and the mean composition of the slag was determined. Table 3 shows the values for the main components of the slag and most important components in terms of viscosity and cooling rate.

**Table 3.** Average composition and liquid temperature of the slag.


The slag material used in further investigations was individually sampled from di fferent positions of the three cooling zones. One sample (H) originated from zone 1, the outer region of the ladle with a high cooling rate. The two other samples (M and S) were sampled from more central areas of the slag inside the ladle, whereby sample M was taken from zone 2 and sample S was taken from the most central zone 3. All taken samples from the designated cooling zones were analyzed by X-ray computer tomography (XCT) and SEM-based image analysis (MLA).

#### *3.1. Results of Computer Tomography*

Results of XCT analysis are given in Figure 8. In the top row of Figure 8, the projection images of the three different samples H, M, and S are shown to give an overview of the structure within the measured volume. The second row shows the cross-sections through the scanned volumes. The red lines visible in the projection images indicate the position of these cross-sections within the volume. The grayscale represents the X-ray attenuation, which is a function of the materials' average atomic number, the wavelength of the X-ray, the material density, and the material thickness. Increased absorption by a region results in a brighter appearance in the image, with pores appearing as dark regions.

**Figure 8.** Results from XCT of samples H, M, and S. Top row: projection images of the analyzed volumes giving an overview of the structures and the size scale; bottom row: reconstructed cross-sections from the position of the red line in the projection images.

The morphological analysis of the metal-rich phases (bright regions in the images of Figure 8) showed that they appear as plate-like to needle-like shapes. Their smallest dimension was often below 100 μm. Mechanical liberation by comminution will, therefore, be challenging, and it requires grinding at least down to an upper particle size of roughly the above given number. The visualization of the metal-rich phases shown in Figure 9 gives an idea of their morphology.

Comparing the projection image of sample H and a slice through the reconstructed volume shown in the first row of Figure 8 demonstrates a problem related to two-dimensional analysis. Pores and spherical strong absorbing phases are clearly visible in the projection image but lacking in the chosen section. Therefore, the strong absorbing phases were analyzed in order to measure their size distribution and volume percentage. This helped to estimate whether or not results from subsequent two-dimensional analysis may be misleading. Figure 10 shows a rendering from the inclusion analysis. Spherical inclusions up to approximately 700 μm in diameter were found. These inclusions only accounted for ~0.4% of the sample volume and can, therefore, be neglected with respect to the processing of the slag.

**Figure 9.** Visualization of the morphology of the metal-rich phases of sample S.

**Figure 10.** Rendering of spherical metallic inclusions contained in sample H.

#### *3.2. Results of SEM-Based Mineralogical Analysis*

In addition to the analyses via XCT, the material was analyzed with MLA according to Section 2.2. Based on these MLA observations, it was shown that the particles in sample H had a different habitus than the particles of the samples M and S, which is in good agreemen<sup>t</sup> with the results of the XCT analysis. The Cu-rich phases displayed a variety of shapes, i.e., forming melting droplets and drop aggregates, veins, or dendritic crystals. Those various Cu-containing phases were present in all samples and had to be compared based on concentration, content of Cu, structure, and recoverability by various methods. The main reason for the previous grinding of the sample to <500 μm is explainable by the anisotropic structure of the investigated sample. Another reason is the available size for a typical grain mount, where bigger particles would cause an insufficient representativity and, at the same time, would lead to a disproportional measurement effort. Furthermore, it would help ge<sup>t</sup> first insight into the liberation behavior of the valuable material.

An overview of the particles analyzed via MLA of the different samples is shown in Figure 11.

**Figure 11.** Overview of the three slag samples H, M, and S: MLA false color images display the difference between sample H and samples M and S.

From the false color images in Figure 11, it can be seen that the elemental content and structure of the particles of sample H were significantly different from samples M and S. The samples M and S can be considered as similar regarding particle composition and properties.

The mass recovery of the size fractions gives an overview of the different breakage behavior for the various slag samples. As displayed in Table 4, the mass recoveries for H, M, and S showed significant differences. The fine fraction <63 μm increased from H to M and to S. The coarsest fraction >250 μm showed the inverse trend.

**Table 4.** Mass recovery of the different size fractions of the samples H, M and S.


The breakage characteristics of the slag seemed to change between the outer and the inner region. More and more fines appeared below 63 μm. One explanation could be the finer intergrown structure of the samples from the inner regions, compared to the mainly homogeneous structure of sample H. Furthermore, the decreasing brittle character of the inner sample led to more abrasion. Nevertheless, the results showed the effect on the breakage characteristics of the cooling rate, which could, therefore, be adjusted to a certain desired behavior.

As Cu was the main valuable metal with significant content within the investigated slag samples, various aspects regarding Cu are discussed in the following sections. It has to be stated that Cu was present in two different forms in the slag material: Cu-bearing slag phase (Cu-rich) and pure metal (Cu-metal). This statement is evaluated in more detail in the following section with various representations focusing on this aspect. The total content of Cu in the three different samples is shown in Table 5. The Cu content is shown for the different size fractions, and an average value is indicated.

Table 5 illustrates that M and S showed a much higher Cu content than H. The Cu content of H was in the range of 3 wt. %, whereas M and S showed approximately 18 wt. % Cu. Furthermore, there was no enrichment of Cu in certain size fractions, resulting in a particle size-independent content. Additionally, the Cu present as pure metal droplets is shown in Table 6. The portion of Cu present as metal represents the amount of Cu that is recyclable in a certain way. The rest of Cu that is present as a slag phase is very likely lost or difficult to gain for a further recycling process. Fortunately, the content of Cu-metal increased from H over M to S. Sample S showed the highest content of Cu-metal with again no enrichment in a specific size class being visible according to Table 6.


**Table 5.** Total Cu content of the different size fractions and average value of samples H, M, and S.

**Table 6.** Cu-metal content of the different size fractions and average value of the samples H, M, and S.


Another aspect of interest concerning a potential recycling process for Cu is the size of these metal droplets within the sample. Therefore, their size distribution is shown in Figure 12. The size distributions of M and S were in the same range, while H showed a much finer size distribution. Nevertheless, all three size distributions were in a rather fine range for a potential mechanical separation process. Furthermore, it needs to be stressed that the required liberation size is around 1/10 of the grain size assuming random fracture [27]. Slags are known to be close to a pure random fracture due to their brittle behavior caused by amorphous phases and rough interfaces [18]. Upgrading Cu-rich phases by mechanical separation processes will, thus, be challenging with respect to the required liberation size.

**Figure 12.** Cu-metal grain size distribution of the three different sampling areas.

A further important characteristic is the shape of the Cu-metal droplets within the different slag structures. From the results of XCT analyses of Section 3.1, it was already concluded that the structure of sample H was different from that of samples M and S. For further investigations, representative examples of the particles analyzed by MLA for sample H (see Figure 13) and for samples M and S (Figure 14) are analyzed in more detail. The shape of the Cu grains in sample H showed a circular shape. In connection with 3D analysis via XCT, it can be concluded that these Cu-metals in sample H were spheres formed by droplets. These spheres (brown color in Figure 13) showed no connection to the surrounding amorphous phase. The amorphous phase in sample H shown in Figure 13 is

represented by the pink color, and it consisted mainly of the elements Ca (~12%), Al (~15%), Fe (~2%), Cu (~2%), Si (~27%), and O (~40%).

**Figure 13.** Typical particles with copper inclusion of sample H: (left, **A** and **C**) grayscales from BSE images; (right, **B** and **D**) associated false color images indicating different phases: pink (amorphous Al slag phase), brown (Cu-metal).

These metal droplets in sample H originated from the turbulent mixing during the smelting process. The metal droplets were transported to the slag phase. The cooling rate in this range was high enough to inhibit settling of the Cu droplets back to the metal phase at the bottom of the furnace. This theory is further confirmed as the Cu droplets showed a sharp boundary to the surrounding slag phases of sample H. A very different observation can be made in samples M and S displayed in Figure 14. In these samples, particles containing Cu comprised Cu-metal droplet aggregates in association with Cu-bearing slag phases.

There are two main areas visible in the false color image of Figure 14. Firstly, there is a more or less homogeneous, needle-like green region (the dark phase in Figure 8 and on the left side of Figure 14 A,C). The second main area is more heterogeneous, marked by brownish colors showing a variety of different phases. The Cu-rich phase consisted of a matrix phase consisting of phases b and c, a dendritic structure (phase d), and the Cu-metal visible as round droplets (see phases in Table 7). The main elemental contents of these dominant phases are shown Table 7. The Cu-metal phase (phase e in Table 7) always showed an agglomeration of spherical droplets within the Cu-rich matrix phases (b and c). The phase d indicated a dendritic structure with a quite high Cu content of approximately 49%. In the whole sample (M and S), the Al-rich (a in Table 7) phase summed up to approximately 50%, the Cu-rich phase summed up to approximately 35% (sum of phases b–d), and the Cu-metal represented 9%.

**Figure 14.** Typical particles with different phases of the samples M and S: (left, **A** and **C**) grayscales from BSE images; (right, **B** and **D**) associated false color images indicating different phases: green (Al-rich slag phase), brownish colors (Cu-rich phases) (see Table 7).


**Table 7.** Main elemental content of the dominant phases of samples M and S.

The Cu-metal droplets always appeared within the Cu-rich matrix (b and c). This fact supports the hypothesis of the occurrence of a transportation process in connection with a formation of Cu-rich dendrites (phase d) and an agglomeration of Cu droplets (phase e). These processes are in fact a function of the cooling rate. In conjunction with the increasing proportion of Cu-metal droplets with more distance to the outer region of the slag (see Table 6), the cooling rate seemed to influence the appearance of the Cu-metal droplets and the characteristics of the surrounding phases.

The deportment of the element Cu, which is the major valuable element in the present material, into the different phases of the investigated slag structures is an important factor for the assessment of its recyclability. Figure 15 shows the deportment of Cu into the different main phases of the three samples H, M, and S.

**Figure 15.** Deportment of Cu as a percentage in the three different samples (H, M, and S) between the different main phases: Cu-metal, Cu-rich, and amorphous.

For sample H, the majority of Cu was found in the amorphous phase. Only 14% of the total Cu content within this sample appeared as the metallic phase. This ratio underlines the low potential for the recycling of the material of this outer zone (H). In contrast to that, the ratios for sample M and S were much more promising with regard to potential recycling, e.g., via mechanical separation processes. For these samples, the amount of Cu present as the metallic phase was much higher (42% and 47%, respectively). The Cu-rich phase in Figure 15 summarizes the phases b, c, and d from Table 7.

For a potential mechanical processing of the material to enrich the Cu-metal phase via separation processes, the liberation of this phase is of major interest. Therefore, in Figure 16, the particle-based information from MLA is binned into property classes of particle size (*<sup>x</sup>*P in μm) and liberation of the Cu-metal phase (*L* in vol. %, indicating the proportion of Cu-metal in a particle). In the individual bins of *x*P and *L*, the theoretical recovery of Cu-metal phase (*r*Cu-metal) is indicated by number and the color scale. The focus of these considerations is only on samples M and S, due to the low potential of sample H for further recycling steps.

The recovery values (*r*Cu-metal) of the Cu-metal phase in Figure 16 summed up to 100% for each sample. According to Figure 16, the distribution in the different *x*P and L classes of the Cu-metal phase was more or less the same for samples M and S with only minor differences. The majority of the Cu-metal was in the range of 40 μm to 320 μm, showing a rather low liberation (≤50%). This material needs to be further liberated by additional comminution steps. These comminution steps have to be controlled (e.g., via combination with a suitable classification process) to avoid an overgrinding of the Cu-metal phase, not to increase the amount in the lower particle size range. Already more than 20% of the Cu-metal is in a range below 40 μm. Despite some fractions showing a satisfying liberation (*L* > 70%), this relatively low size range is already challenging for conventional mechanical separation processes (e.g., density separation via shaking tables).

**Figure 16.** Theoretical recovery (*r*Cu-metal) of the Cu-metal phase into size (*x*P) and liberation (*L*) classes for the two main products M and S.

#### *3.3. Other Elements*

The above evaluations focused on the element Cu, due to its relatively high content. There are other interesting elements present in the investigated material. Table 8 shows the averaged content of some other interesting elements in samples M and S.

**Table 8.** Averaged content of various elements in samples M and S measured by MLA.


The appearance and deportment of the elements of Table 8 are decisive for the potential recycling of these materials. For samples M and S, the majority of these elements were distributed between the Cu-rich phases (mainly phases b, c, and d of Table 7) and the Al-rich phases (mainly phase a of Table 7). Some elements were present in other phases with significant amounts. Table 9 gives an overview of the deportment of these elements within samples M and S.

**Table 9.** Deportment of various elements between the Cu-rich, Al-rich, and other phases in samples M and S.


Obviously, the majority of Al was found in the Al-rich phases; nevertheless, 29% was found in the Cu-rich phases, and 1% was present as Al metal inclusions. For Fe, the main amount was found in the Cu-rich phases, whereas only 21% was in the Al-rich phases, and 20% was present as a more or less pure oxide phase. The situation for Sn was totally different, as no Sn was found in the Al-rich phase, while 30% was present in the Cu-rich phases, and the majority appeared as isolated SnO2 in the sample. Nearly all Pb was in the Cu-rich phases. For Zn and Ni, the majority was found in the Al-rich phases. In contrast to Cu, the other elements had no significant appearance in a pure metallic form, while some

minor metallic occurrences were found only for Al (1%). This analysis of the deportment of various elements gives a first overview of the di fferent behavior of these elements. A more detailed evaluation of these elements and other trace elements requires the application of additional analysis methods [28].
