Superduplex stainless steel is a widely used material in many industrial areas, due to its significant properties in terms of mechanical and corrosion resistance. In order to guarantee the quality of these steels, the ferrite evaluation is an important analysis since many properties depend on the control of the ferrite content. During equipment manufacturing or field inspections, the most conventional way to evaluate the ferrite content is through ferritoscope measurements, due to its portability and easy interpretation. However, it has been observed that the calibration pattern used for ferritoscope calibration can lead to inaccurate ferrite quantification in superduplex inspection. In order to analyze this influence, different characterization techniques were performed in superduplex samples, such as optical microscopy, X-ray diffraction, and vibrating sample magnetometer, to compare with the measurements obtained by a ferritoscope. It was concluded that for high ferrite values, with the calibration made from the calibration pattern provided by the ferritoscope manufacturers, the ferrite values measured showed significant deviations from the real values. The current work presents a solution for using the ferritoscope for a more accurate quantification of the ferrite content in superduplex stainless steels.
The evaluation of ferrite content in superduplex stainless steel is an important step in the manufacturing process of many industrial equipment. An accurate evaluation contributes for the project life-span prediction, avoiding premature failures or unscheduled stops. International standards establish limits for ferrite content in stainless steel, for instance, NORSOK M-630 [1] defines between 35 and 55% of the acceptable ferrite content for the base metal of duplex and superduplex stainless steel, and a range of 35–65% for the welded parts. In the case of austenitic stainless steels, a value from 2 to 5% of ferrite content in the weld bead is not only acceptable, but also expected [2], ensuring a higher resistance to stress corrosion cracks and reducing the risk of solidification cracking [2,3]. Different techniques of microstructural characterization are commonly applied to evaluate the ferrite content in stainless steel, such as optical microscopy, scanning electron microscopy, X-ray diffraction, metallographic replication, vibrating sample magnetometer, and the ferritoscope [4–7]. Among the listed techniques, it is reasonable to emphasize the ferritoscope as the most appropriate technique for the ferrite content evaluation in industrial environment, because besides its portability, the equipment is simple to operate and provides a direct non-destructive response for different types of stainless steels [8]. To use the ferritoscope, it is mandatory to calibrate the equipment with calibration pattern defined by the standard AWS A 4.2 [9]. The known ferrite amount in the calibration samples is correlated with ferritoscope magnetic response, and a specific calibration curve is configured in the equipment. Considering the guide presented by [10], it is introduced a procedure and a set of calibration samples to experimentally measure the ferrite content in austenitic and duplex and superduplex stainless steel. However, the experimental results achieved by the present study following the suggested calibration procedures did not achieve significant results inspecting superduplex stainless steel. In addition, it is highlighted through [9] that materials with the same volumetric fraction of ferrite can present different magnetic behaviors, depending on the grain size and manufacturing process. The current work made a careful analysis of the ferrite content results achieved by the ferritoscope and presents a specific solution for using the equipment for a more accurate quantification of ferrite in superduplex stainless steels.
2MethodsTwenty-three samples of superduplex stainless steel class UNS S32750 with chemical composition presented in Table 1 were prepared. Different heat treatments were performed in the samples with the aim to produce different amounts of ferrite. The samples have 90×45×5.5mm, and the ferrite volumetric fraction was quantified by X-ray diffraction (XRD).
XRD was performed in a D8 Discover (Bruker AXS) using cobalt radiation (λ=1.789Å), equipped with a Lynx Eye PS Detector. The equipment operated at constant values of tension (35kV) and current (40mA), respectively. The voltage and current values were 35kV and 40mA, respectively. The scanning data was obtained in the 2θ range of 45–105°, with a step size of 0.001° and scanning velocity of 0.5s/step. Ferrite volumetric fraction was obtained by the Rietveld method [11], using the Diffrac Plus TOPAS® software. The ferrite volumetric fractions were measured in nine different areas of each sample and the ferrite average value was calculated as presented in Table 2. The XRD technique has been widely used for quantitative analysis of different phases in multiple-phase materials [12], and therefore, it is the choice for such task in this work. The ferrite average values achieved by XRD in the superduplex samples were defined as reference values to compare with the ferritoscope measurements. Fig. 1 shows an XRD spectrum of the phases presented in sample 16 that had an average volumetric fraction of ferrite (δ) and austenite (γ) around 49.7 and 50.3%, respectively.
Twenty-three superduplex reference samples were prepared with different heat treatments to achieve several amounts of ferrite. The ferrite contents were evaluated by XRD.
Sample | Ferrite content (%) | Sample | Ferrite content (%) |
---|---|---|---|
01 | 32.5±2.7 | 13 | 52.2±2.0 |
02 | 47.5±3.5 | 14 | 55.7±4.9 |
03 | 34.8±4.1 | 15 | 52.8±1.6 |
04 | 45.1±11.5 | 16 | 49.7±7.7 |
05 | 31.7±7.8 | 17 | 43.1±5.1 |
06 | 29.6±8.2 | 18 | 45.7±5.7 |
07 | 36.6±4.9 | 19 | 61.1±2.9 |
08 | 33.4±4.5 | 20 | 71.6±5.0 |
09 | 41.2±6.9 | 21 | 55.1±3.0 |
10 | 40.4±5.3 | 22 | 58.2±6.9 |
11 | 39.0±7.1 | 23 | 65.7±6.3 |
12 | 29.9±3.6 |
Analyses by optical microscopy were also made in some samples to corroborate the XRD results. The samples were grinded, polished, and electrolytic etched with 20% NaOH solution, applying 3V for 15s to reveal the microstructure. Fig. 2 shows the metallographic image of the samples 16 and 20, with 49.7 and 71.6% of ferrite, respectively.
The magnetic induction ferrite meter used in the work was the FERITSCOPE® FMP30 from Helmut Fischer® with the calibration pattern WRC 0.3/80, which contain samples with 0.52, 3.05, 10.2, 30.4, and 55.1% of ferrite and one sample, named as ferrite base, containing 100% of ferrite. It is important to mention that the ferritoscope test is mandatory to calibrate the equipment with the calibration pattern [10]. Moreover, to study the calibration pattern samples, the following equipment were used: X-ray fluorescence spectrometry for chemical composition analysis with Bruker's TRACER III-SD; vibrating sample magnetometer (VSM) from Lake Shore to analyze the material magnetization curve; and Elcometer® 456 to evaluate the presence of metallic coatings.
3Results and discussionsThe first approach was to calibrate the ferritoscope with the calibration samples provided by the manufacturer and test the equipment efficiency measuring the ferrite content in all 23 superduplex samples prepared. The ferrite values indicated by the ferritoscope were correlated with the ferrite measured by XRD. Fig. 3 shows the ferritoscope ferrite content as a function of the values measured by the XRD. The line on the graph indicates the theoretical agreement between the techniques. As noticed in the graph, the ferrite values are underestimated by the ferritoscope. However, it can be highlighted that the samples with low amount of ferrite presented a better correlation with the XRD measurements and as the ferrite content increases the correlation is impaired.
The results shown in Fig. 3 raise the hypothesis that the calibration curve set on the ferritoscope might have a good fit just for low ferrite content. In order to investigate the results, a detailed analysis of the standards samples used in the ferritoscope calibration process was performed. Through the X-ray fluorescence technique, it was estimated the chemical composition of calibration samples, where chromium (Cr), nickel (Ni), molybdenum (Mo), and iron (Fe) contents are shown in Table 3.
Among the specimens used in the ferritoscope calibration, it is possible to emphasize three samples that showed unexpected characteristics on the first investigation. On the chemical composition results, the standard sample with 55.10% of ferrite presented a Cr content of 96.76%, and, due to this high amount, it was suspected that the sample presented a metallic Cr coating. An evaluation of the metallic coating was performed using the equipment Elcometer 456, which revealed a Cr coating of 25.5μm of thickness. A mechanical grinding removed the coating, and a new chemical composition analysis was made, revealing a carbon steel with 0.095% Cr, 0.007% Ni, 0.39 Mo, and 96.96% Fe. The ferrite base specimen, who corresponds to 100% of ferrite, also presented an anticorrosive metallic coating. The third specimen that presented an unexpected result was the one with 30.40% of ferrite, because considering its chemical composition (Table 3), it corresponds to a duplex stainless steel, which, as annealed, presents a balanced microstructure with the ratio of ferrite/austenite around 50/50% [13,14]. However, according to the calibration pattern, the amount of ferrite in the sample is 30.40%. The first investigation to confirm the microstructure presented in the 30.40% calibration specimen was the optical microscopy. In this analysis, the specimen was mechanically grinded and polished, followed by electrolytic etching. Fig. 4 shows the duplex steel 2205 microstructure of the specimen used in the ferritoscope calibration pattern, which the phase counting was performed according to the standard ASTM E562 [15]. The sample presents 51.4% of ferrite, differing from the 30.40% described in the calibration standard.
Besides the metallographic analysis, aiming to confirm the ferrite content in the sample, a magnetization curve of the sample was performed, which, according to Abreu et al. [16] and Silva et al. [17], from the magnetic saturation level is possible to establish the content of the ferromagnetic phases presented in the material. A vibrating sample magnetometer was used to trace the magnetization curve of the calibration sample that indicates 30.40% of ferrite and the curve of the as annealed sample, number 16 (Table 2) with 49.7% of ferrite. These curves were compared in order to verify the similarities of the saturation levels. As presented by Fig. 5, the saturation level of the samples is close, indicating a similar ferrite amount. It is worth mention that the difference at the beginning of the magnetization curves is caused by the different geometries of the samples and to avoid such variation is necessary to apply different magnetization factors in the samples which were not considered in the curves. However, to evaluate the ferromagnetic content, the saturation level is sufficient, and it does not depend on the samples geometry [18], thus validating the similar ferrite content in the samples.
Based on the chemical composition analysis, optical microscopy, and the magnetic saturation level, it is possible to conclude that the calibration sample that indicates 30.40% of ferrite actually has a volumetric fraction of ferrite close to the as-annealed situation, with approximately 50% of ferrite. This difference in the quantification might be one of the reasons for the underestimated result achieved by the ferritoscope showed in Fig. 3. The ferrite content evaluation is a polemic subject, widely discussed and detailed by [9]. In order to improve the results of the ferritoscope quantification, three samples of the 23 prepared in this work, were selected to replace some samples of the ferritoscope calibration pattern. Table 4 shows the ferrite content and the chemical composition of the samples selected for the new calibration of the ferritoscope, samples number 12, 16, and 20 (Table 2) replaced the standard calibration samples with high ferrite. These samples were applied to calibrate the ferritoscope and later all the superduplex samples were again tested. It is important to note that the sample with 29.9% of ferrite also presents intermetallic phases, such as sigma, in the microstructure. The superduplex phase stability diagram and the influence of the sigma phase in the superduplex magnetic behavior are well reported in the work published by Bettini et al. [19] and Rebello et al. [20].
Fig. 6 shows the ferrite content measured by the ferritoscope as a function of the ferrite reference values achieved by XRD. Comparing Figs. 3 and 6, it is possible to notice that the ferritoscope calibration with the superduplex samples improved the ferrite quantification. The results confirm the relevance of using a calibration standard sample with the same material as the one intended to inspect. Consequently, to achieve accurate ferrite quantification in superduplex steels, it is necessary to calibrate the equipment with samples of the same material under inspection, which differs from what is stated by [10].
4ConclusionAccording to the results, for stainless steels with low ferrite content, until 35% of ferrite, it is possible to use the calibration pattern provided by the ferritoscope manufactures. The fact that the ferritoscope calibration pattern is composed of different alloys lead to an underestimation of the ferrite content in dual phases material, such as superduplex. Results showed that for ferrite quantification in superduplex stainless steels, it is necessary to use a specific standard with the same material and same manufacture process for the ferritoscope calibration.
5Conflicts of interestThe authors declare no conflicts of interest.
The authors would like to thank the Brazilian research agencies CAPES and CNPQ for the financial support.
Paper was part of technical contributions presented in the events part of the ABM Week 2017, October 2nd to 6th, 2017, São Paulo, SP, Brazil.