The search for increasingly cleaner steels has heightened the demand for additional analysis techniques, especially for the evaluation of inclusions in steel where greater cleanliness is required. A range of factors should be taken into account when selecting a particular method, in accordance with analysis objectives and in order to maximize the reliability of results. Although statistical techniques make it possible to correlate data from smaller samples with entire heats of steel, some methods are more suited to evaluating different inclusion profiles. The objective of this study was to evaluate the main characteristics of certain techniques used to study inclusions. Two of the primary methods for direct inclusion analysis of solid steel are metallographic techniques and chemical analysis, with total oxygen content used as an indirect inclusion measurement. A search of the literature identified the main advantages and disadvantages of each method, as well as the primary limitations for their use. This makes it easier to determine the most suitable methods for carrying out the desired analysis.
Non-metallic inclusions cause modifications in the properties of steel (for example, in the fatigue limit); however, for some applications, the occurrence of inclusions of a certain size may result in catastrophic failures [1–5]. When non-metallic inclusions diminish the production capacity or properties of the steel in service, it is no longer considered a clean steel and requires inclusion control to ensure quality [6]. Inclusion control should specify how the inclusions are distributed in the material and the size ranges in which they are found, as well as obtain the morphologies and chemical compositions that give rise to less damaging inclusions [7]. The foremost concern is to prevent the formation of inclusions that compromise the material in question or facilitate their removal when they emerge and, subsequently, to alleviate the effects using the appropriate treatment [7,8]. Understanding the behavior of inclusions during steel processing is essential in order to ensure that the final product complies with client specifications and quality standards. The most common form of detecting harmful inclusions, in line with specific quality parameters, is by characterizing them. This can be done in a number of steel processing stages [9–11], depending on the factors to be analyzed. The aim of the present study is to determine the efficiency of scanning electron microscopy and total oxygen measurements in characterizing inclusions, as well as the difficulties involved in employing these methods.
2General aspects on the characterization of inclusionsThe characterization of inclusions consists of evaluating their size, distribution, morphology and chemical composition (type of inclusion). There are a variety of techniques that can be used industrially to determine the inclusion content of steel. However, it is important to consider the size of inclusions, time needed for inspection and the volume of the material to be analyzed [12–14]. According to Hénault [12], Fig. 1 shows the detection limit in accordance with the mass of the material to be analyzed using different methods. Fig. 1 also shows that the lower the ultrasound frequency the greater the volume of material that can be analyzed. However, this increases the minimum size of the defect that the technique is capable of detecting.
![Detection limit in accordance with the mass of the material to be analyzed using different methods [12].](https://multimedia.elsevier.es/PublicationsMultimediaV1/item/multimedia/S2238785415000277:gr1.jpeg?xkr=ue/ImdikoIMrsJoerZ+w9/MYgIby5y4uSkdCdnNKiOmycX6/4J/OS+LK7YOmtWVpQbB7GIHuiYagTvLV1jxNSgVJwpNh9FPgfqsVvjC2mY9R4boKeV6ix5iFT1GeovnwhTpjpFa1U7nUR+BAuBx82s9r4T9kMl48DGMOQ8qYKHJzpJae3gL+JVuHnQqIJWuJFCEY11EdcAssmMiFH8C1vSiqZHC80YNtDpBrrTyW4RReXQQnT7Dsf6SPJ+qDkGtPbWp6ERjVW1J8DVVTP6ziP2nxd877QQEEsblpyrsK9KM= "Detection limit in accordance with the mass of the material to be analyzed using different methods [12].")
Detection limit in accordance with the mass of the material to be analyzed using different methods [12].
Fig. 1 indicates that when the volume of the material requiring analysis increases, the minimum size of the detectable inclusions also increases. This limits the use of ultrasound methods for identifying inclusions in steel where inclusion control involves smaller inclusions, such as bearing steel. Thus, the use of smaller representative steel samples is becoming increasingly common for analysis methods that encompass smaller inclusions in their detection range. Indeed, more than one technique is needed to characterize the total inclusions population. Methods based on the surface of the sample identify inclusions with a maximum size of 20μm, sufficient to characterize the population of micro-inclusions. Methods based on sample volumes, for inclusions larger than 20μm, detect inclusions that can cause potentially catastrophic effects during manufacturing processes or at the beginning of the life cycle of a steel part. Despite the limitation of assuming the behavior of a homogeneous heat, due to the small sample size [12], statistical methods aid in the treatment of data so that these better correspond to the reality of the entire heat [13].
According to Zhang [14], there are several methods that allow the detection of inclusions in volumes of solid steel as well as sections of solid and liquid steel. These ranges from conventional methods of like optical microscopy to ultrasound testing, optical emission spectroscopy, laser induced spectrometry, X-ray computed tomography and automated scanning electron microscopy [14,15]. For steel with low inclusion content and narrow permissible size ranges, methods with a satisfactory detection limit are necessary. Among the techniques most widely used to analyze micro-inclusions is scanning electron microscopy coupled with energy-dispersive X-ray spectrometry (SEM/EDS), which simultaneously provides an elementary chemical analysis and metallographic images. As an additional improvement, more modern equipment includes an automated system, which enables the entire sample to be scanned [16]. Indirect methods are also widely adopted in industry to determine the number of inclusions and consist of determining the chemical composition of slag, the conditions of refractory parts and even the levels of elements dissolved in the bath [14]. A commonly used technique involves measuring the total oxygen content of steel samples, which is directly related to the presence of inclusions [14,17]. Most of the equipment employed for this method simultaneously measures the dissolved nitrogen in the sample, enabling nitrogen pickup to be analyzed. The difference in dissolved nitrogen between samples from different stages of the process may be indicative of air absorption by the bath [14], which is directly related to reoxidation.
3Sample collection for analysisSample type is a key variable in correctly evaluating the occurrence of inclusions and affects any analysis method adopted. Table 1[18,19] shows the most common sample types and different characterization techniques that can be used, considering both direct and indirect methods. At times, a technique that may be used on a certain type of sample is not recommended because of difficulties associated with sampling or the analysis technique.
Sample types and possible techniques applied [18,19].
| Type of sample | Analysis | Characterization techniques |
|---|---|---|
| Immersion samplers (steel) | Chemical composition of steelInclusions analysisTotal oxygen | Optical Emission Spectroscopy (OES), Pulse Discrimination Analysis (PDA), Oxygen AnalyzerSEM/EDS |
| Pin (steel) | Chemical composition of steel (carbon and sulphur)Inclusions analysisTotal oxygen | Optical Emission Spectroscopy (OES), Pulse Discrimination Analysis (PDA), Oxygen AnalyzerSEM/EDS |
| Bar; Plate (steel) | Chemical composition of steelInclusions analysisTotal oxygen | Optical Emission Spectroscopy (OES), Pulse Discrimination Analysis (PDA), Oxygen AnalyzerSEM/EDS |
| Slag | Chemical composition | X-ray Fluorescence |
| Clogging Mass | Chemical composition and Morphology of inclusions | SEM/EDS |
Immersion samplers are common in industry since they can be used both for SEM and total oxygen determination. Table 2[20–23] shows some of the types of samplers and possible protection methods adopted.
Immersion samplers used in industry [20–23].
| Sampler | LP-6 | LP-12 | BB | TOS | Samp-O-line |
|---|---|---|---|---|---|
| Thickness | ∼6mm | ∼12mm | ∼14mm | ∼5mm (Ø) | ∼12mm |
| Metal-cap-protect | Yes | Yes | Yes | No | Yes |
| Argon-protect | Yes | Yes | Yes | Yes | Yes |
| Figure | |||||
LP, lollipop; BB, bjöneborg; TOS, total oxygen sample.
Important parameters to be considered when choosing a sampler are penetration, depth and immersion time, since these determine whether the sampler is completely filled with steel. Moreover, these parameters directly affect the cooling rate of the sample, which influences the analysis. According to Ericsson [23], the number of secondary inclusions in liquid steel samples increases significantly with the rise in the cooling rate of the samples. The protection adopted may also interfere in the results of inclusion characterization. In general, argon protection is more recommended because samples containing deoxidizers are more easily contaminated by slag, which modifies total oxygen measurements [22]. Furthermore, deoxidizers can alter the chemical composition of inclusions in the steel sample [17]. Precautions taken with sampling depend on the factors and type of steel to be analyzed. The total oxygen sampler (TOS) prevents the entry of air and has the added advantage of minimizing sample preparation, since sectioning is not necessary [21]. However, use of the TOS is limited owing to its size, which restricts the detection of larger inclusion populations. Each type of sampler offers specific advantages and drawbacks and satisfactory results should encompass the reliability of data to be measured, compatibility with the techniques adopted and cost. Fig. 2 shows an example of using the total oxygen technique to assess cleanliness throughout the processing of CS18 steel in the furnace [5]. The vertical bars depict the mean and standard deviation for each phase of the process. Total oxygen values fall sharply along the samples, as desired.
![Evolution of total oxygen and dissolved oxygen for samples of CS18 steel heats [5].](https://multimedia.elsevier.es/PublicationsMultimediaV1/item/multimedia/S2238785415000277:gr2.jpeg?xkr=ue/ImdikoIMrsJoerZ+w9/MYgIby5y4uSkdCdnNKiOmycX6/4J/OS+LK7YOmtWVpQbB7GIHuiYagTvLV1jxNSgVJwpNh9FPgfqsVvjC2mY9R4boKeV6ix5iFT1GeovnwhTpjpFa1U7nUR+BAuBx82s9r4T9kMl48DGMOQ8qYKHJzpJae3gL+JVuHnQqIJWuJFCEY11EdcAssmMiFH8C1vSiqZHC80YNtDpBrrTyW4RReXQQnT7Dsf6SPJ+qDkGtPbWp6ERjVW1J8DVVTP6ziP2nxd877QQEEsblpyrsK9KM= "Evolution of total oxygen and dissolved oxygen for samples of CS18 steel heats [5].")
Evolution of total oxygen and dissolved oxygen for samples of CS18 steel heats [5].
The information required in order to characterize inclusions includes their size, distribution, morphology and chemical composition. Among the possible direct methods, SEM results meet these objectives, resulting in their widespread adoption by industry and laboratories. Nevertheless, total oxygen measurement offers the advantage of speed and simplicity in identifying inclusions in a sample already used in SEM.
4.1Manual and automated SEMOne of the most commonly applied techniques to measure steel cleanliness is by automated SEM equipped with an EDS probe. The result is a semi-quantitative analysis of the chemical composition of inclusions on a surface prepared using conventional metallography, as well as the position, shape and size of inclusions [15]. One of the primary advantages of applying the automated method is the possibility of obtaining data for all the particles scanned in the sample [7]. Manual methods are limited to analyzing only those inclusions detected by the operator of the equipment. Moreover, there is no guarantee that the sample will be assessed equally and in its entirety. Automated SEM enables equal scanning of the sample and analyzes all the identifiable inclusions. Analysis time depends on SEM characteristics, the extent of automation, analysis parameters, sample area assessed and the cleanliness of the sample itself, among others.
The classification of inclusions in automated methods is a useful tool; however detailed knowledge of the filter system used is needed. Given that the analysis is considered semi-quantitative, results may be distorted. This is the case, for example, when it is necessary to distinguish between the different calcium aluminates formed as a result of calcium treatment in aluminum-killed steel. Table 3 shows the variation that can occur in the adjustment of SEM parameters based on different studies cited in the literature. One of the major limitations of the method is related to the representative area of the sample. Due to the combination of low inclusion occurrence in clean steel and the limited volume of the sample for cleanliness assessment, the meaning of the results is heavily dependent on the size of the area analyzed [13,25]. Another limiting factor is the size of the inclusion; in general, the larger the particle the smaller the steel matrix around the inclusion [8,11,14]. As such, quantifying the composition of micro-inclusions similar in size to the electron volume interaction (beam) is problematic and often inaccurate [28]. Most studies specify a cut-off inclusion size of 1μm for automated analysis (Table 3).
Parameters adjusted in SEM.
| Author | Material | Area (mm2) | Beam energy (kV) | Working distance (mm) | Minimum particle size (μm) | Analysis time per particle |
|---|---|---|---|---|---|---|
| Bartosiaki et al. [7] | SAE 52100 | 62 | 20 | 16–18 | 4 | Min. 1sMax. 2s |
| Kaushik and Yin [18] | AHSS steel | 150–180 | 20 | – | – | – |
| Michelic et al. [13] | Stainless steel | 100–200 | 15 | 10 | 4 | 3s |
| Kaushik et al. [24] | Steel | 180 | 20 | 15–20 | 1 | 2s |
| Graham and Yang [25] | HY-100 | >30 | – | – | 1,1 | – |
| Verma et al. [26] | Laboratory: deoxidized Al–MgIndustry: AlKLC | – | 20 | – | – | – |
| Nuspl et al. [27] | Low carbon | – | 15 | 18–20 | 1 | (EDX) 3s |
Data obtained via automated analyses can be more reliably represented in distribution graphs, since a much larger population of inclusions can be considered in the sample. Chemical analysis of inclusions enables the chemical composition to be depicted in ternary phase diagrams, using database management systems [7,29,30]. Limits of this analysis include normalization of the chemical composition of inclusions in ternary and pseudo-ternary systems, which may mask the real classification of the inclusion. In addition, for automated techniques, the chemical composition of scanned inclusions considers the mean of the chemical elements detected in each inclusion, therefore disregarding the occurrence of different phases in the inclusion itself, such as the agglomeration of different compounds, which is typical of exogenous [31] or oxysulfide inclusions [32]. An efficient means of characterizing inclusions is by combining manual SEM performed by an operator with automated SEM analysis. This allows the general behavior of all the inclusions to be mapped and provides a more detailed analysis of specific inclusions of interest. Another drawback of both methods is the conversion of chemical elements into non-metallic compounds, since the adoption of different criteria for the classification of inclusions produces different results. Classification criteria can be anything from simple separation between oxides and sulfides to filtering criteria for separation between different types of oxides (for example: alumina, calcium aluminates, calcium silicates, etc.). Moreover, the results obtained via manual SEM vary not only in accordance with the parameters of the device used and sample preparation, but also with the operator's level of experience.
4.2Total oxygenTotal oxygen content is widely used to characterize steel cleanliness in terms of oxide inclusions and assess process improvement and quality control. Total oxygen is the sum of soluble oxygen in liquid steel and that present as oxide inclusions [7,14,17]. The major advantage of this technique is the ease and speed of obtaining values that are directly related to cleanliness and easier to correlate with meltshop process parameters, such as processing times and the chemical composition of the steel itself and slag.
The most common method for measuring total oxygen is a testing procedure using equipment that simultaneously analyzes oxygen and nitrogen in the sample [18,19]. Although this is a quick and easy technique for evaluating cleanliness during the various production stages; it is only applicable for oxide inclusions and provides no information on the chemical composition and morphology of inclusions. However, when data from this technique are used in conjunction with the results of SEM–EDS testing, for example, the final analysis is far more detailed. Industrial samples may exhibit significant variations in OTotal values. As mentioned in item 3, the choice of sampler and sample preparation techniques must be consistent in order for the obtained results to be compared and correlated with the process conditions of industrial heats [18]. This is a critical factor in analyzing steel with high internal cleanliness, where maximum total oxygen values do not exceed 0.0010%. The limit value is that of thermodynamic equilibrium, which is a function of the chemical composition of steel, as discussed by Holappa [33]. For aluminum-killed steels, for example, the minimum value will be between 3 and 5ppm [33].
5ConclusionsConclusions reached in regard to tools for characterizing inclusions are as follows:
- •
Regardless of the analysis to be performed, the type of sample should always be selected as a determining criterion in obtaining representative results.
- •
The manner in which sampling is carried out also affects results, since penetration, depth and insertion time determine whether the sampler is correctly and completely filled with liquid steel.
- •
The selection of a method should take into account a number of factors, including the type of sample and the techniques used to analyze it, as well as the type of steel to be evaluated.
- •
Techniques that involve larger sample volumes analyzed in a viable time encompass larger inclusions in their detection limits and, as such, are not recommended for identifying micro-inclusions.
- •
With regard to the comparison between manual and automated SEM, automation enables data to be obtained for all the detectable inclusions in the sample scanned, ideal for distribution analysis, for example. However, certain peculiarities, such as phases and morphologies, can be better assessed using the manual method, which analyzes each detail according to the operator's specifications.
- •
Total oxygen measurements can be quickly and easily obtained. Nevertheless, these measurements only consider oxide inclusions and provide no information on the inclusions present in the sample. The best way to use this analysis is combine it with other techniques.
- •
The main advantage of using total oxygen measurements is their efficient correlation with process data, since they represent the micro-inclusion population.
- •
While measurements of total oxygen scale the number of inclusions found in the sample, electron microscopy combined with chemical analysis enables the types of inclusions present to be characterized.
- •
Whenever possible, a combination of techniques is recommended in order to ensure the most comprehensive characterization of inclusions possible.
The authors declare no conflicts of interest.
The support of CNPq, CAPES, FINEP and UFRGS is gratefully acknowledged.
