2018 Volume 58 Issue 6 Pages 1079-1085
To understand the γ→α transformation and control the carbon contents among the phases involved is most important in the design of high-strength and high-ductility low carbon steels. Using a previously-developed field emission-electron probe microanalyzer (FE-EPMA) which is able to suppress hydrocarbon contamination to a very low level, the carbon concentrations at γ/α interfaces and within the austenite were analyzed in Fe-xC-2.0Si-yMn (x=0.15 or 0.20, y=1.5 or 2.0 in mass%) steels isothermally transformed at 750°C and 800°C. The paraequilibrium (PE) model gives a good accounting of carbon enrichment in the 1.5 mass% Mn steel held for 15 s, whereas the NPLE/PLE transition model of local equilibrium gives a much better accounting in the 2.0 mass% Mn steel than PE. However, the interfacial carbon concentration agrees with the composition shown by the NPLE/PLE transition line in both alloys when annealed for 1800 s. In the 1.5 mass% Mn steel, the carbon concentration at the interface in austenite seems to have shifted from PE to NPLE during annealing.
In recent years, a combination of higher strength and improved ductility has been demanded in high strength steel, which is used in automotive parts. These properties are achieved by using hard phases such as martensite and bainite to increase strength and strain-induced transformation of the retained austenite phase to improve elongation.1,2,3) When using martensite, bainite and retained austenite to improve the properties of steel, it is important to understand the relationship between the microstructure and carbon distribution with high accuracy and high spatial resolution.4,5,6,7,8,9,10)
The steel microstructure is formed through phase transformation. Since the phases involved contain carbon and substitutional alloy elements which differ greatly in diffusional mobility, much research has been carried out on the composition of the phases involved under guiding principles such as local equilibrium (LE)11,12) and paraequilibrium (PE).13) Using ternary Fe-C-X and quaternary Fe-C-X1-X2 alloys, electron probe microanalysis (EPMA) was performed to measure the distribution of the alloy elements in large matrix grains in specimens after extended isothermal holding, presumably after the partition local equilibrium (PLE) mode began.
Recently, the early stages of the γ→α transformation were examined in Dual Phase steel by using field-emission gun (FE)-EPMA.14,15) Analysis of the early stages is important for understanding the relationship between the carbon distribution and the steel microstructure. However, quantitative microanalysis of carbon with an electron probe is quite difficult because the electron-beam irradiated area on the specimen surface becomes contaminated in the analysis process.
To address this problem, we developed an FE-EPMA equipped with various anti-contamination devices. These devices can maintain a lower limit of quantification of carbon as low as 0.01 mass% even in mapping analysis. In contrast, conventional devices are limited to 0.1 mass% in point analysis.16) In the following, this device is called the Carbon analyzer (C-analyzer).
In this paper, we measured the carbon distribution in Fe-0.15C-2Si-1.5 and 2Mn alloys which were austenitized and isothermally held for a short time. Based on a comparison of the measured results with the results of a DICTRA simulation,17) the carbon concentration at the ferrite/austenite interface and the equilibrium condition at the growing ferrite/austenite interface are discussed.
EPMA is conventionally used for quantitative chemical analysis of steel specimens. However, it is not suitable for quantitative analysis of carbon because carbon contamination accumulates on the sample surface during measurements. Contaminants, such as hydrocarbons present on the sample surface and in the vacuum chamber, interact with the incident electron beam and cause carbon to accumulate. Previously, this type of carbon contamination was decreased by pretreating the samples by polishing and/or installing a liquid nitrogen cold trap directly above the sample in the measurement chamber. However, with these conventional devices, carbon can only be analyzed with an accuracy of 0.1 mass%, even in point analysis.
The method of measuring carbon in the steel microstructure by using our system is non-destructive and requires that the analysis area be smaller than the matrix grain size. FE-EPMA is suitable for analyzing such a small area in a non-destructive manner. The system utilizes the following three contamination reduction and suppression techniques.
1. Plasma cleaners and liquid nitrogen traps are installed in the sample chamber and sample preparation chamber to remove hydrocarbons from around the sample.
2. A heating stage is used to maintain the sample temperature at 100°C so as to suppress hydrocarbon accumulation around the sample during measurement.
3. Three carbon analyzing crystals are used to reduce measurement time by increasing sensitivity.
While sample contamination is reduced by plasma irradiation of the sample in scanning electron microscopy (SEM) and transmission electron microscopy (TEM), this technique is not suitable for EPMA because irradiation in the measurement chamber damages the organic membrane window on the detector tip. We placed retractable protective plates in front of all detectors for protection from plasma irradiation, but in spite of this measure, contamination still accumulated as the electron beam irradiation time increased. To avoid this, we devised an original heating stage using a ceramic heater which operates at around 100°C during EPMA measurements. Figure 1 shows a schematic of the developed instrument equipped with the above-mentioned contamination reduction devices.
Schematic diagram of developed instrument for carbon mapping of steels.
Table 1 shows (a) the carbon contents determined by chemical analysis, (b) the average carbon contents determined by FE-EPMA, (c) the difference between the results in (a) and (b) and (d) the standard error of 16 measurements by FE-EPMA for each specimen. These results show not only high accuracy but also very good inter-date repeatability of the point measurements with the aid of the plasma cleaner in the specimen chamber and the liquid nitrogen trap. Under the experimental conditions used in this study, the carbon content at a selected point on the sample can be determined with accuracy of 0.01 mass% or more and inter-date repeatability of about 0.01 mass% in the range of 0.089 to 0.46 mass% C.
The alloys were vacuum-induction melted using high purity electrolytic iron, high purity carbon, manganese and silicon. The ingots were hot-rolled and cold-rolled into 1 mm thick steel plates. The chemical compositions of the alloys are shown in Table 2. As schematically illustrated in Fig. 2, the samples were austenitized at 950°C for 10 min and isothermally held at 750°C or 800°C for 15 s or 1800 s using fluidized bed furnaces, and then quenched in water. The L cross-section of the samples was mirror-polished for quantification of C, Si and Mn by the C-analyzer.
Schematic diagram of heat treatments.
The measurements of carbon were performed first with a focused beam diameter of ~50 nmϕ at an accelerating voltage of 7 kV and a probe current of 50 nA. Then, line analysis and quantitative mapping analysis of carbon were carried out under conditions that minimized the accumulation of contamination. The details of this procedure have been described elsewhere.18) Under the measurement conditions used in this work, the spatial resolution of carbon analysis was 100 nm or less, and the calibration curve method with standard samples of Fe–C alloy was used to quantify carbon. The measurements of Si and Mn were then performed in the same field of view at an accelerating voltage of 9 kV and a probe current of 100 nA in order to reduce the signal-to-noise ratio in the analysis of the K line. In the samples used in this work, Si and Mn are not expected to partition between ferrite and austenite. Hence, after this assumption was confirmed with one sample, Si or Mn were not analyzed in the other samples of each alloy.
Furthermore, in order to observe the microstructure of the same area as that measured by the C-analyzer, the samples were lightly polished and etched with 3% nital. The fine details of the microstructure were observed using in-lens (matter accentuation) imaging at an accelerating voltage of 1 kV by FE-SEM (Supra55VP, Carl Zeiss).
To analyze the phase transformation behavior, the program DICTRA (ver. 2015b) and the TCFE7 and MOB2 databases were used. The latter is a database of the diffusion coefficients in multi-component systems evaluated by the CALPHAD method.19)
Figure 3 shows the DICTRA calculation model used for the calculations. Figure 3 schematically illustrates the diffusion cell used in the simulation. The cell shape was rectangular, and the α phase (1 × 10−9 m thick) was set to grow from the left side of the γ phase cell. The length of the entire cell was set to 25 μm or 10 μm, which are half of the maximum and minimum grain sizes, respectively, to consider variations in the matrix grain size. In addition to local equilibrium, growth under the paraequilibrium mode was also simulated at 750°C and 800°C.
Calculation conditions of phase transformation in DICTRA.
Figures 4(a) and 4(b) show the microstructure observed by FE-SEM at a low accelerating voltage and the quantitative carbon map measured by the C-analyzer, respectively, in the case of alloy A held at 750°C for 15 s. Figure 4(c) is the line profile of carbon taken horizontally across the area (indicated by an arrow). It can be seen that the α phase did not divide the austenite matrix, the carbon concentration in the central area of the martensite matrix is close to that in the bulk (0.15 mass%), and carbon is enriched in front of the interface, while in some regions near the interface carbon is not enriched, which may indicate that the whole interface did not move uniformly.
a) SEM image(accelerating voltage=1 kV), b) quantitative mapping and c) line scan of carbon in alloy A isothermally held at 750°C for 15 s.
Figures 5(a), 5(b) and 5(c) show the microstructure, the carbon map and the line profile of carbon in the same sample, but in a different area in which the matrix is surrounded by impinging α grains to form an isolated island. These figures indicate that, while a large γ grain had the same concentration as in the bulk (0.15 mass% C), carbon was enriched to a larger extent in small γ grains. Thus, the degree of carbon enrichment differs according to the location in the microstructure and/or along the interface, even in a sample heat-treated under the same conditions.
a) SEM image(accelerating voltage=1 kV), b) quantitative mapping and c) line scan of carbon in alloy A annealed at 750°C for 15 s.
Figures 6(a), 6(b) and 6(c) show the microstructure, the quantitative carbon map and the carbon line profile for alloy B held at 750°C for 15 s. The proportion of the γ phase was higher in this alloy because of a higher carbon content. The mapping image indicates that a large amount of carbon discharge occurred in the region where a large amount of the α phase was formed. The higher carbon concentration at the interfaces indicates that the interfaces were moving.
a) SEM image(accelerating voltage=1 kV), b) quantitative mapping and c) line scan of carbon in alloy B annealed at 750°C for 15 s.
Figures 7(a) and 7(b) show the low magnification quantitative carbon maps obtained by the C-analyzer for alloys A and B held at 750°C for 15 s. The carbon concentration at the center of the large γ grains was close to the bulk concentration, which indicates that the C-analyzer measured a difference as small as 0.05 mass% in alloys A and B without being affected by contamination. It is noteworthy that, even in alloy A with the lower carbon content, a large amount of carbon enrichment occurred in the small γ grains during the ferrite transformation. Thus, two-dimensional visualization of the carbon distribution in the microstructure has been realized, which was not possible with the conventional technology.
Quantitative carbon mapping of a) alloy A and b) alloy B annealed at 750°C for 15 s.
Figures 8(a) and 8(b) show the evolution of the α phase fraction and carbon profiles, respectively, simulated at 750°C in alloy A with the cell size of 10 μm. Although the Mn and Si profiles are not shown, these elements were not partitioned at the holding time of 1800 s, which is consistent with the results of the measurement by the C-analyzer. Figures 8(c) and 8(d) show the evolution of the α phase fraction and carbon profiles simulated with the cell size of 25 μm. When the cell width or the austenite grain size was narrower, soft impingement of the carbon diffusion fields started to occur earlier. As a result, the carbon concentration at the center of the γ grain increased more rapidly.
a) Evolution of α phase fraction and b) carbon profiles simulated at 750°C in alloy A calculated with cell size of 10 μm, and c) evolution of α phase fraction and d) carbon profiles with cell size of 25 μm in alloy A.
Figures 9(a) through 9(d) show the carbon profiles after holding at 750°C for 15 s, together with those of alloy A held at 800°C for 15 s (Fig. 9(b)) simulated under the LE and the PE modes.†) The carbon concentrations at the interface in austenite were higher under the PE mode than those under the LE mode regardless of the cell width.
Calculated C profiles in isothermal annealing of a) alloy A at 750°C for 15 s, b) alloy A at 800°C for 15 s, c) alloy B at 750°C for 15 s and d) alloy C at 750°C for 15 s with local equilibrium and paraequilibrium.
Next, we attempted to determine whether LE or PE dominates in carbon partitioning by analyzing the carbon concentrations at a relatively early stage of transformation. Since the scatter of the carbon concentration in the matrix is small (see Fig. 9) and moreover, the interfacial carbon concentration did not depend on the matrix grain size, we believed that it might be possible to determine the dominant transformation mode even if the spatial resolution of C-analyzer was not sufficient. For this purpose, quantitative mapping measurements were performed in five areas at the same magnification as in Figs. 4, 5, 6 for alloy A held at 750°C and 800°C for 15 s and alloys B and C held at 750°C for 15 s. More specifically, line data were scanned at five to seven locations in which the α phase was formed. The maximum carbon concentrations observed at the interface as well as the carbon concentrations in the matrix were obtained from each line profile at more than 30 locations in total.
In low carbon steel, it is considered that the ferrite transformation proceeds under PE in the very initial stages and then switches to the no-partition local equilibrium (NPLE) mode and finally the partition local equilibrium (PLE) mode. Indeed, Hutchinson et al.21) reported that the growth rate of proeutectoid ferrite in an Fe–C–Ni alloy suddenly slowed several tens of seconds after the transformation started. The initial fast growth was ascribed to the absence of a Ni diffusion spike, implying that the growth occurred under a condition close to PE, and after growth slowed, the diffusion spike built up, which indicates that the transition to the NPLE mode occurred. In order to determine the transformation mode, it is necessary to calculate the composition at the interface under the LE and PE modes. Figures 10(a) and 10(b) show the interfacial tie-lines for ferrite growth under LE and PE in an Fe–C–Mn–Si quaternary alloy. Point a indicates the intersection with the α/(α + γ) phase boundary of a carbon component ray (straight line from the carbon apex of the tetrahedral phase diagram with constant composition ratios of Fe, Mn and Si) that passes through the bulk composition O. Point c is the other end of the equilibrium tie line which started from point a. While carbon activity is represented by a line in a ternary system, it takes the form of a curved plane in a quaternary system. Point b, i.e., the intersection of the carbon component ray with the carbon isoactivity plane, represents the carbon concentration at which the transition from the NPLE mode to the PLE mode occurs. Under PE, the interfacial tie line (d e) coincides with the carbon component ray. Thus, the NPLE/PLE and PE boundaries and the transition temperatures can be calculated following the same procedure as in a ternary system. Although this alloy contains 2.0 mass% Si, which is greater than the Mn mole fraction, the influence on the path of the α/γ phase boundary and the NPLE/PLE boundary is smaller than that of Mn.13,22)
a) Local equilibrium (ac) and b) paraequilibrium interfacial tie-line (d e) for growth of ferrite in Fe–C–Mn–Si alloy (O). O’ is the projection onto the basal plane or Fe–C–Mn phase diagram.
Figures 11(a) and 11(b) show the carbon concentrations at the NPLE/PLE and PE boundaries together with the T0 composition for alloys A and C calculated using the TCFE7 database. In the 1.5 mass% Mn alloy, some data points reached the PE boundary concentration at both 800°C and 750°C at 15 s (Fig. 11(a)). In contrast, in the 2.0 mass% Mn alloy (Fig. 11(b)), all data indicate that the carbon concentration at the interface is near the NPLE boundary after holding for 15 s.
Comparison of measured carbon concentrations with PE phase boundary, PLE/NPLE boundary and T0-line calculated from Thermo-Calc in isopleth a) at 2.0 mass% Si-1.5 mass% Mn and b) at 2.0 mass% Si-2.0 mass% Mn.
The carbon concentrations at points c and e in Fig. 10 are both quite high. Point c of the local equilibrium interfacial tie-line is the carbon concentration at the diffusion spike of Mn. Assuming that the ferrite thickness in the sample with the holding time of 15 s is around 10 μm (see Fig. 4 etc.), the average growth rate was v ≈ 0.6 μm/s. Since the diffusion coefficient of Mn in austenite is D~10−18 m2/s,22) the Mn spike width w is calculated to be
In Fig. 12, the carbon concentrations obtained by the C-analyzer are compared with the DICTRA results in samples held for 1800 s. From the left to right, the data point for the lowest carbon content is obtained from alloy C held at 800°C, followed by alloy A held at 800°C, and the rightmost point is from alloy B held at 750°C. The points near the origin were taken from the measured and calculated carbon concentrations in the α phase. The results of measurement by the C-analyzer are in good agreement with those of the DICTRA calculation, which indicates that the growth of the α phase occurred under LE at around 1800 s.
Comparison of carbon content between measurement by C-analyzer and calculation by DICTRA.
As seen above, in the 1.5 mass% Mn alloy, the ferrite transformation occurred initially (t < 15 s) under PE, and the transformation proceeded under LE at around 1800 s. In the 2.0 mass% Mn alloy, the transformation proceeded under NPLE at around 15 s at 750°C. Thus, in this work, the transition from the PE mode to the NPLE mode in the ferrite transformation was shown for the first time in terms of the carbon concentration at the interface, which was made possible by high-precision carbon quantification using the C-analyzer.
These findings are important in intercritical annealing in actual work processes. For instance, in the short Continuous Annealing Line (CAL) process, where a constant temperature is held for about 120 s, the above results indicate that the rate of the γ → α transition is altered significantly by small differences in the amount of added Mn. Since the transformation rate is crucial for deciding the industrial process conditions, it is very important to understand as precisely as possible the effect of the Mn content on the kinetics of phase transformation.
In previous work, the authors developed an FE-EPMA called the C-analyzer, which is equipped with anti-contamination devices and can perform elemental analysis with high precision of 0.01 mass%. In the present research, the carbon concentrations near the interface were compared with the results of DICTRA calculations in the early stages of the ferrite transformation in quaternary Fe–C–Mn–Si alloys. The results are as follows.
(1) In an Fe-0.15 mass% C-2.0 mass% Si-1.5 mass% Mn alloy, the transformation was likely to proceed under the paraequilibrium mode at around 15 s during holding at 750°C and 800°C, while at around 1800 s, the transformation mode changed to the local equilibrium mode (NPLE and PLE).
(2) In an Fe-0.15 mass% C-2.0 mass% Si-2.0 mass% Mn alloy, growth under paraequilibrium was not observed around 15 s, indicating that the local equilibrium mode dominated during holding at 750°C.
These results demonstrate that the transformation rate depends sensitively on the Mn content and affects the design of process conditions. It is also important to develop a method for simulating the two- or three-dimensional distribution of elements in the microstructure.
This work was the result of a project of the Innovative Structural Materials Association (ISMA) and was supported by the New Energy and Industrial Technology Development Organization (NEDO).