Influence of steel cleanliness on drawability of fine filaments with high tensile strength

The aim of the current study is to clarify the influence of steel cleanliness on the fracture during wet wire drawing of high strength high carbon very fine filaments. Methodologies for inclusion identification and for inclusion count were developed. Ninety percent of the fractured filaments were associated with an inclusion which was SiO2, or SiO2-CaO or SiO2-MgO. The fracture of the wires resulted from debonding of the inclusion from the matrix. Inclusion density assessed from total oxygen content in the wire was successfully and reliably obtained. Inclusions exhibited a wide range of sizes but mostly around 6mm. The inclusion density did not appear as the only parameter that affects the fracture occurrence. An important effect of size and of chemical composition of the inclusions was found. A critical size of inclusions for fracture was observed but it depends on the chemical composition of inclusions.


Introduction
Non-metallic inclusions play an important role on crack initiation and further fracture of metallic alloys subjected to mechanical loading. This can occur during service of the material as well as during processing. Respectively, fracture associated with non-metallic inclusions has been observed when the material is under fatigue [1][2][3] or when it is wiredrawn or rolled [4]. Though these two situations strongly differ in the loading level, it is shown that the presence of a single inclusion can be detrimental enough to lead to fracture. Attention is today much paid on the role of inclusions on fatigue resistance especially in very high cycle fatigue. Unfortunately, it is very hazardous to transfer directly the acquired knowledge in the field of fatigue to the field of wire drawing or cold rolling. Indeed, these processes involve huge amount of monotonic plastic deformation which can result to very thin sheet or very small diameter of wire.
For some applications (filters, saw wires, wire ropes...), superfine steel wires are needed. The market requires very fine wires with very high mechanical properties. However, fractures might occur during the wet wire drawing process that leads to 60 mm filaments. The main types of fracture during drawing fine filaments are central bursting, tension break and fracture due to inclusions. The main feature of fine filaments is that the size of the inclusion approaches the diameter of the wire so that only one single inclusion can be responsible for the fracture. Very fine wires are currently processed on tungsten [5], gold [6] and copper [7][8][9]. Both process parameters (drawing speed, die angle…) and inclusions properties (shape, size...) were found to influence the drawability. Steel wires are more complicated to be studied because of the complexity of the microstructure being generally multiphased. Some studies shows the impact of inclusions on steel wire for bigger diameter by finite element analysis (FEA) [10,11]. Nevertheless, the identification of inclusion characteristics is a prerequisite for the understanding of failure mechanism of very fine wires. The characterization of inclusions includes chemical, mechanical and morphological aspect [12,13]. The quality of the results depends on the number of analyzed inclusions which extraction is easier in bulk material than in fine wires.
The present paper deals with the fracture of very fine wires of pearlitic steels with diameter less than 80 mm. It aims first at pointing out the importance of the presence of inclusions. Then it focuses on the determination of the inclusions characteristics (size, chemical composition, shape) and on their impact on fracture occurrence during the fine filament drawing.

Materials
The investigated filaments have been drawn from wire rods of 5.5 mm in diameter provided by different suppliers. The typical chemical composition is reported in Table 1, the carbon content of the wires varies from 0.94 to 1.02%C. It is assumed that carbon content only influences mechanical properties but not the inclusion composition and their density.
The cleanliness quality of the steel was dependent upon the steelmaker. It is defined by the density and the composition of non-metallic inclusions present in the wire.
Rods of 5.5 mm were transformed by several steps of dry wire drawing and heat treatment. During the final step called wet wire drawing (WWD), the intermediate wire was drawn into filaments of diameter down to 60 mm by more than 20 passes.

Experimental procedure
For assessment of the inclusion morphology, the Dekker extraction method [14] was employed in order to dissolve the matrix and to recover the inclusions. The chemical solution contained a mixture of HNO 3 , H 2 SO 4 and H 2 O in the respective proportion 1%:38%:61%.
All filaments that fractured during WWD were collected for fracture surface observations and chemical analysis of the inclusions by scanning electron microscopy (SEM). This was performed on a FEI Quanta 200 3D SEM coupled with the Oxford Instruments EDS system and operating at an accelerating voltage of 15 kV.
The size of the encountered inclusions was determined by means of equivalent circular diameter (ECD) as defined by equation (1). For calculation hypothesis, the inclusions were considered with an ovoid shape: where S inc is the surface inclusion defined by equation (2): Here, r L is the radius of the length of the inclusion and r w the radius of the width of the inclusion.
To define the morphology of inclusions, the aspect ratio of the inclusions is defined as follows: Aspect ratio ¼ r L =r w : An aspect ratio of 1 corresponds to a spherical inclusion and higher is the value, the more elongated is the inclusion.
For the investigation of the cleanliness of the steel, automated SEM/EDS analysis of polished longitudinal section specimen were performed using a JEOL 7200F FEG-SEM and an Oxford Instruments EDS system. The automated inclusions measurement with SEM/EDS provided the size of inclusions, the number of inclusions, and the chemical composition of the inclusions. Inclusion density was calculated as follows (Eq. (3)): where d inc is the density of inclusions per/cm 2 , n is the number of detected inclusions, S total is the total scanned area.
In addition to the SEM/EDS technique, the cleanliness was also evaluated by oxygen measurement using a combustion technique with a LECO TC500. Indeed, the total oxygen content can be a reasonable indirect way to evaluate the total amount of oxide inclusions in steel [13].
3 Results and discussion

Fracture cause
More than 200 fractured filaments during WWD were investigated in order to determine the fracture cause. Fracture surfaces of filaments either contained an inclusion (Fig. 1a) or were free of inclusion (Fig. 1b).
An inclusion was present in more than 90% of the fractures. Although this ratio appears much higher than that reported in the literature [7] for copper (52%), it is closer to the one observed for inclusion related fracture on bigger diameter steel wire (80%) [10]. Inclusions into fracture surface are always in the centre of the wire and on the top of a cone fracture.
When an inclusion was not involved in the fracture of the filament, the latter exhibited a slight necking and the fracture surface contained dimples. Therefore, though the filament was strongly deformed during the drawing process, it kept a certain degree of ductility.

Identification of inclusions 3.2.1 Chemical composition
EDS analysis were performed directly on the inclusions emerged from the fracture surfaces (Figs. 2a-2c). Three major types of inclusions were identified on the fracture surfaces.
All the analysed inclusions have been identified as Sibased inclusions. The latter were pure silicon oxide (SiO 2 ) (Fig. 2a) or a more complex composition SiO 2 -CaO ( Fig.  2b) or SiO 2 -MgO (Fig. 2c). CaO and MgO refer to the second major compound present into the inclusion.
Si-type inclusions appear to be non-deformable during wire drawing of fine filaments [15]. Figure 2 shows also the different morphologies of inclusion responsible for fracture. The SiO 2 inclusions appeared to be rather round while SiO 2 -MgO appeared to be rectangular with sharp angles. The shape of SiO 2 -CaO inclusions was difficult to appreciate by this technique.
In order to evaluate the inclusions morphology as well as their effect on the surrounding matrix, sectional analysis of the fractures was carried out (Figs. 3a and 3b). For both cone and cup part of the fracture, it appears that some   defects parallel to the drawing axis are present around the inclusion, however no real crack opening are observed along the drawing axis. Some inclusions seem to be more prone for debonding. The presence of certain chemical elements is known to play a role on debonding of the inclusion into the fracture. Here, this aspect is confirmed as shown in Figure 4. Hence, the aluminium-rich inclusion (Fig. 4a) appears more prone to debonding than the alumium-free one (Fig. 4b). This statement is in correlation with the study of where the aluminium content is proved to be promoting the debonding matrix/inclusion during fatigue test [16].

Size
The identified inclusions also differed in their ECD as well as in their quantity as shown in Figure 5. The size of inclu sions involved in filament fracture is ranging from 1.90 to 23 mm but the majority of inclusions exhibits an average of ECD of 6 ± 3 mm.  A link between the composition and the critical size of inclusions found on fracture surfaces is observed. The chemical composition is expected to influence the critical size of the inclusion for drawability by different ways. By means of FEA, Yu et al. [17] pointed out cracks near inclusions after rolling as a result of a non-uniform deformation around the inclusion which was emphasized with increasing the inclusion size. As well, the inclusion size and inclusion composition impact the fatigue properties by changing the local stress state in and around the nonmetallic inclusions as a result of the strong difference in Young modulus between the inclusion and the matrix [11].
Although Si-inclusions are known to be non-deformable during forming process [15], secondary major elements also exhibit an effect in regards to the WWD process. Hence, the critical size of the inclusions involved in the fracture is dependent of the CaO and MgO presence. Indeed, CaO decreases the critical size of the inclusions leading to fracture, whereas MgO tends to increase it. Figure 6 shows the amount of fractures related to each type of inclusions. SiO 2 -MgO inclusions are the most frequently encountered (51%) while SiO 2 inclusions seem to be less dangerous for fracture occurrence (11%). Furthermore, variations in hardness between these inclusions are expected.
However, even if the SiO 2 -CaO inclusions are responsible of 38% of the fractures, a statistical analysis of the distribution of the inclusions within the wire rod (Tab. 2) indicates that this type of inclusions is only present on 17% of all the inclusions present into the wire.

Factors promoting fracture 3.4.1 Inclusion density
The first parameter suspected to have an influence on fracture occurrence was the inclusion density. Indeed, as it is claimed that the cleanliness of steel wire has an impact on the drawability of wires, the relationship between inclusion density and fracture is one of the points to consider. Here, the drawability has been evaluated by counting the number of fractures encountered per 1 ton of processed material. When the number of fracture per ton exceed a predefined threshold (Th inc -1 ton )*, the material was considered bad in term of drawability. Otherwise, the material is considered good for processing. In order to pre scan the inclusion population, the cleanliness was assessed from LECO oxygen measurement. The LECO device measures the total oxygen contained in the steel while SEM/EDS provides a density of inclusions containing oxygen. Figure 7a shows that there is a simple relation between the total oxygen content (O tot ) and the oxide density. Hence, a reliable link between the oxide density and the measurement of the total oxygen content O tot relation can be established.
Comparing drawability and oxide density (Fig. 7b) points out a general tendency. An exceptional drawability is observed for very low oxide density (about 50 inclusions per cm 2 ). For higher oxide densities, the scattering does not allow establishing a clear correlation. The wires classified as bad in terms of drawability contained an oxide density below 400 inclusions per cm 2 . However, even with this density or higher (up to 600 inclusions per cm 2 ), some wires had satisfied the qualification. Therefore, the oxide density estimated by the measurement of total oxygen content provides a first idea of the global drawability. It is encouraged to decrease the oxide density in the steel for decreasing the amount of fracture per ton during the WWD. However, one must have in mind that there is no clear oxide density threshold below which will result in a good or bad drawability. In conclusion, the oxide density is not the only factor that has an influence on the number of  fractures per ton. The drawability depends also on inclusions properties such as the size, the morphology and the chemical composition.

Inclusion size effect
Both fractured and non-fractured filaments have been investigated in order to consider if one of the inclusion features appeared to be the most relevant for the fracture. Fine filaments were embedded and polished so that the longitudinal section could be observed. One hundred inclusions in a non-fractured filament were examined by SEM. It was found that there was no preferential inclusion in terms of chemical composition in the fractured and nonfractured wires. All the types of morphologies were found in fractured and non-fractured filaments. However, it has been found that the average size of inclusions in the non-fractured wires is much lower than that in the fractured filaments (Fig. 8). Indeed, it turns out that there is a critical size of about 3 mm for the inclusion to cause the fracture of the filament. Independently of the chemical composition, inclusions lower than 3 mm are considered as not dangerous for WWD of 60 mm filaments.
This implies that not only the quantity on inclusions present into the wire is important but also the size of the inclusions.

Effect of inclusion shape
Beside the chemical characterisation of the inclusions, their morphologies have also been assessed.
The precipitates retrieved from the dissolved matrix were observed by SEM to obtain the 3D morphology together with the composition. SiO 2 inclusions have a rounded shape with an aspect ratio equal in average to 1.05 which translates a spherical shape. (Fig. 9a), SiO 2 -CaO inclusions (Fig. 9b) present an ovoid shape and an aspect ratio average of 1.26. While SiO 2 -MgO inclusions are mainly rectangular with sharp angle and elongated with an aspect ratio of 1.37 (Fig. 9c). The identification of the inclusions with the extraction method confirms and defines the morphology of inclusions assessed on fractures. SiO 2 -MgO particles exhibit sharp angles, suggesting therefore they could act as stress concentrators [18] and accelerate the process of fracture. However, this was not just the case.

Conclusion
The present study aims at linking the fracture of very fine filaments of pearlitic steels with the presence of inclusions and leads to the following conclusions: inclusions are responsible of fractures for more than 90% of the case during WWD; three main types of inclusions have been found: SiO 2 , SiO 2 -CaO and SiO 2 -MgO; the fracture of the wires results from the debonding of the inclusion from the matrix; the inclusion density does not control alone the drawability but also the composition, size and morphology of inclusions; the chemical composition of inclusions plays a role on the critical size of inclusions for fracture.   Radius of the length of the inclusion (mm) r w Radius of the width of the inclusion (mm)