1 Introduction

Hot forming in combination with die quenching is widely used in the production of high-strength components because of the high-strength-to-weight ratio.[1] The knowledge about the phase transformations occurring during the hot forming process is essential for the design of a component with targeted microstructural and mechanical properties.[2] Depending on the chosen process parameters, as for example the austenitizing temperature or cooling rate, different phase transformations take place, which define the final microstructure. The time dependent phase transformations are illustrated in isothermal (TTT) or continuous cooling transformation (CCT) diagrams.[3]

Phase transformations can be detected due to changes in lattice structure and specific volume.[4] They can be recognized as deviations of the cooling curve as a result of their exothermal nature.[5] Standard methods for monitoring phase transformations in steels are dilatometer tests, differential thermal analysis (DTA) or differential scanning calorimetry (DSC).[6]

A dilatometer can obtain data with a variety of measurement techniques such as linear variable differential transformers, extensometers or laser. The specimens were heated by induction heating, restive heating, radiation or convection in furnaces.[7]

In modern dilatometers, flow stresses, phase transformation kinetics, and the approximation of the thermal expansion coefficient can be determined.[8] Different heating and cooling strategies as well as different strain rates can be performed to simulate the hot forming process. The temperature change, measured by a thermocouple, and the global specimen dilatation during the whole thermo-mechanical treatment are recorded and the phase transformations are visualized with dilatation curves.[8]

A thermo-mechanical simulator, which is for example provided by Gleeble, is also used to simulate the influence of the hot forming process on the resulting microstructural and mechanical properties.[9] Dilatation curves can be generated similar to a dilatometer test.[8]

Using DTA, the temperature differences are measured during heating and cooling of the specimen in relation to a thermally inert material.[10] Phase transformations can be monitored as deviations from the zero line by a change in heat content and thermal properties of a specimen.[11]

DSCs can measure heat flow rates as well as characteristic reaction temperatures. Compared to the DTA technic, the difference between the heat flow rate of the specimen and a reference sample are measured during a regulated temperature profile. Due to thermally activated reactions the specimen temperature undergoes an alteration, which led to a DSC signal.[12]

For the determination of the final microstructure for relevant thermo-mechanical process routes, normally a dilatometer or a thermo-mechanical simulator is used in combination with metallographic analysis and hardness tests. Only one process route can be simulated with one sample, which resulted in a high experimental effort and high costs for the design of load-adapted components with tailored properties.[13] For many steels, continuous cooling transformation diagrams (CCT) are already published in literature, whereas the choice of deformation continuous cooling transformation diagrams (DCCT) is limited. The latter are indispensable for a targeted design of hot-formed components.

To reduce the experimental effort Bambach et al.[13] presented an approach to use a virtual dilatometer for 22MnB5 steel, which can reproduce real dilatometer curves when the model parameters are formulated as a function of the pre-strain. Reitz et al.[14,15] introduced an innovative characterization method for the determination of phase transformations using optical methods, which enables the investigation of the effect of different process parameters on the phase transformation kinetics using one specimen. Digital image correlation (DIC) was performed at high-temperature to determine sample dilatation during cooling, and thermal imaging was used to detect the temperature change. The experimental setup enables high deformation degrees and allows the representation of close-to-process conditions, especially with regard to the cooling conditions.

The use of the DIC technic for the contactless full-field measurement across a specimen enables the determination of specimen deformation and strain within the test region. Pan et al.[16] utilized high-temperature DIC at 1200 °C to analyze the full-field thermal deformation and coefficient of thermal expansion. Further areas of application are full-field 3D measurements,[17] static and quasi-static loading, high-speed dynamic loading.[18,19] The supplier of thermo-mechanical simulators also provide additional equipment for DIC full-field measurements.[9]

The combined investigation of the material behavior by DIC and thermal imaging was also performed by Chrysochoos et al.[20] to determine the mechanical energy and heat sources involved locally during a heterogeneous tensile test at room temperature. Kilic et al.[21] monitored the thermal response of a composite using the same optical methods to analyze failure mechanisms. The calibration of the thermal imaging and DIC pictures are discussed in Reference 22.

The investigation of the phase transformation kinetics for tailored thermo-mechanical processed specimens was introduced first by the authors in References 14, 15, where the resulting microstructure and mechanical properties were compared with literature data. Until this time, the authors were not aware of any studies in this field where DIC in combination with thermal imaging were utilized for the analysis of the transformation kinetics for tailored thermo-mechanically treated flat steel specimens.

Because of the novelty of this approach, the aim of this study is to determine the accuracy of the innovative optical characterization method by a direct comparison with dilatometer tests and already existing dilatation curves in literature. Additionally, a new approach is presented to demonstrate how the temperature distribution of flat specimen can be influenced by a flux concentrator to reach a targeted temperature profile for further investigations of tailored steel processing. Conventional approach is to use numerical simulation to get an optimized shape of an inductor for a targeted microstructure profile as for example introduced by Schlesselmann et al.[23] The advantage of the new approach is that instead of searching for an optimal inductor shape, the principle of adjusting the shape of the flux concentrator is much easier to perform. For this purpose, the final concentrator geometry is adapted to the respective inductor cross-section by means of milling. In areas where a lower temperature is desired, material on the flux concentrator is further reduced, resulting in a localized decrease in the coupling of the magnetic field. This allows a relatively intuitive adjustment of a desired temperature profile within a sample.

The second aim is to demonstrate the field of application and possibilities of the optical characterization method by two examples of a phase transformation analysis of graded thermo-mechanically processed specimens.

2 Experimental Procedure

2.1 Material

The experimental study was carried out using a 22MnB5 steel sheet with an AlSi-coating and an uncoated steel sheet. Both sheets had a thickness of 2 mm. All tensile specimens were cut along the rolling direction of the sheet by water jet cutting. To determine the chemical composition of the flat sheets, spark spectroscopy (Q4 TASMAN from Bruker AXS GmbH) was utilized. The results are presented in Table I. All AlSi-coated specimens were pre-alloyed in a resistance furnace (NABER) at 700 °C for 15 minutes and cooled on free air before further processing and testing. Pre-alloying prevents melting of the AlSi-coating during DIC analysis due to the completed diffusion of iron into the coating and formation of intermetallic phases with a high melting point. This was necessary because otherwise the speckle pattern would be destroyed by the melting of the AlSi-coating when the sample is heated up to the austenitizing temperature.

Table I Chemical Composition of the 22MnB5 Flat Steel Sheets
Full size table

2.2 Characterization Method for the Determination of Phase Transformations

The optical characterization method for the determination of phase transformations, which was employed in this investigation, was introduced by Reitz et al.[14,15]

For the detection of phase transformations, the experimental setup of Figure 1(a) was used. To determine the local change in temperature a thermal imaging camera (VarioCAM® HD head 980) was utilized. A temperature measurement range of 250 °C to 2000 °C was predefined by the calibration of the camera. For the determination of the temperature dependent emission coefficient, comparison measurements were performed with a type K thermocouple. With this data, a script for the emission coefficient was implemented in the thermal imaging software. To detect the local displacement of an area of interest in tensile direction, high-temperature digital image correlation (DIC) was performed using a digital camera (Nikon D3200). The light emitted from the specimen as a result of heating the steel up to the austenitizing temperature led to not evaluable DIC pictures. This problem was solved by the application of a blue bandpass filter as suggested by Grant et al.[24]

Fig. 1
figure 1

(a) Experimental setup for the determination of phase transformations: (1) specimen with speckle pattern, (2) medium-frequency oscillator with frontal inductor in combination with a magnetic flux concentrator, (3) chuck jaws of the hydraulic universal testing machine, (4) air nozzle, (5) ratio pyrometer, (6) thermal imaging camera, and (7) digital camera with blue bandpass filter; (b) specimen for the dilatometer test; (c) specimen for the optical method

Full size image

For the direct press hardening process the blanks are heated to the austenitization temperature, transferred to the press after soaking and simultaneously formed and quenched in water cooled dies.[25] The resulting process parameters are the austenitizing temperature, soaking duration, degree of deformation and cooling rate. Such a process route could be simulated using the experimental setup of Figure 1(a). Due to the connection between the universal testing machine (MTS), the medium-frequency generator (Trumpf) equipped with a controller (Eurotherm), the ratio pyrometer (Advanced Energy) and the air nozzle with a regulating valve it is possible to investigate different austenitizing temperatures, soaking durations, degrees of deformation and cooling rates. Since the specimens in Section III–B were not tested above the uniform strain, the target global strain of the specimen was defined using the machine displacement of the MTS. Through the Eurotherm regulator, the current temperature could be transmitted to the MTS controller via the pyrometer, allowing the power of the generator to be adjusted. The compressed air flow of the quenching nozzle was controlled manually via a valve. The achieved cooling rates for a specific compressed air flow were determined experimentally by quenching tests.

The dog-bone shaped specimen geometry for the dilatometer tests and thermo-mechanical tests of the novel optical characterization method are shown in Figures 1(b) and (c).

To perform the analysis of the phase transformation kinetics using the optical method, the specimens were treated according to the flow chart depicted in Figure 2. First, the specimens were sandblasted, cleaned with ethanol and then painted (Process step 1). A black high-temperature varnish as a prime coat and a white high-temperature varnish for the speckle pattern were applied by spraying.

Fig. 2
figure 2

Flow chart for the phase transformation analysis using the optical method

Full size image

Before the thermo-mechanical treatment took place, the frontal inductor equipped with a surrounding U-shaped flux concentrator was aligned behind the rear specimen surface with a coupling distance of 2 mm. Then the specimens were heated up to austenitization temperatures between 900 °C and 950 °C with a heating rate of 10 K/s, using the high frequency generator with a frequency of 250 kHz (Process step 2). After soaking for 300 or 600 seconds, an optional tensile deformation with a strain rate of 0.025 s−1 was performed. Undeformed specimens underwent the same temperature ramp. Both cameras took pictures simultaneously with a recording frequency of 2 Hz during the end of soaking and cooling of the specimens. The cooling after the optional deformation was performed by compressed air using an air nozzle. Some of the specimens were cooled down to the ambient temperature on ambient air. For the calculation of the cooling rate, the method explained in Reference 26 was utilized.

The obtained pictures of both cameras were analyzed in the software GOM Correlate Professional, which enables the analysis of the change in elongation and temperature (Process step 3). With the help of measured elongation and temperature data, dilatation diagrams were established to detect possible phase transformations.

The described method (illustrated by Figure 2: Process step 2 and 3) for the determination of phase transformations by means of optical methods is called “optical method”.

After the thermo-mechanical treatment and dilatometer tests, all specimens were cut for metallographic analysis and hardness measurements to be examined as described in section 2.6 (Process step 4–7).

2.3 Numerical Modeling of the Induction Heating Process of Flat Steel Specimens

As described in the introduction, this section will present the new approach to obtain a targeted temperature distribution within a flat tensile specimen. The inhomogeneous temperature distribution using induction heating is a disadvantage of this energy efficient heating principle as shown in Figure 3.[27] The application of the flux concentrator is shown in Figure 3(a). For the use of a homogeneous flux concentrator geometry, the maximum temperature is located within the middle of the specimen with a temperature of 810 °C (s. Figure 3(b)). For the expansion of the application potential of the optical method, the specific influencing of the sample temperature was essential, since, e.g., small tensile samples can be taken from homogeneous heated areas as demonstrated in Reference 28 or heating scenarios of tailored heating processes can be simulated. Figure 3(c) demonstrates how the temperature distribution of a tensile specimen can be manipulated to get a linear temperature gradient. For this purpose, a 3 deg angled concentrator geometry was utilized. In Figure 3(d), the two temperature profiles using the homogeneous or angled concentrator geometries are plotted along the line L1. The region of maximum temperature with 815 °C is shifted downward from the center of the specimen by the use of the angular concentrator. A temperature gradient of 55 °C could be reached utilizing the 3 deg angled concentrator.

Fig. 3
figure 3

(a) Scheme of specimen and inductor with flux concentrator (b) resulting temperature profile with a homogeneous concentrator geometry (c) resulting temperature profile with an angled concentrator geometry (d) temperature profile L1 for both concentrator geometries

Full size image

By locally reducing the concentrator material, the coupling of the magnetic field is locally degraded, allowing desired temperature gradients to be set.

To reduce the experimental effort a finite element model was developed for the analysis of the influence of the magnetic flux concentrator geometry on the resulting temperature distribution of the tensile specimen heated by induction.

A coupled simulation in Ansys Workbench® R2020 was implemented to model the induction heating process. With the eddy current simulation in Ansys Maxwell 3D the generated ohmic loss inside the specimen can be calculated. To reduce computational time, a simplified CAD model of the experimental setup, presented in Reference 29, was constructed and is depicted in Figure 4. For the magnetic flux concentrator made of Fluxtrol® 50, a nonlinear relative permeability was considered according to the data sheet of the material.[30] A temperature dependent nonlinear relative permeability for the material card of 22MnB5 was also updated with the data of References 31 and 32. Furthermore, the transfer of the determined ohmic loss into the steady-state thermal analysis system allows the recalculation of the ohmic loss into a heat generation. To include the temperature dependence of the material parameters, the feedback iterator is used in Ansys Workbench®. The coupled simulation is repeated until a target delta-temperature of 5 pct is reached. Heat losses due to radiation and convection are considered in the setup of the steady-state thermal analysis. The temperature dependent coefficient for convection presented in Reference 33 is integrated in the material library of the engineering data for 22MnB5.

Fig. 4
figure 4

Simplified model of the induction heating of a flat specimen using a linear inductor with a flux concentrator

Full size image

Three different flux concentrator geometries were considered. The initial state, which represents the “Geometry 1” depicted in Figure 5, revealed an inhomogeneous temperature distribution visible in Figure 3(b). Only at a length of about 8 mm from the middle of the specimen, a homogenous temperature distribution could be detected. Therefore, the middle area of the flux concentrator was reduced, as seen in Figure 5 (“Geometry 2”), to decrease the temperature in the middle area of the specimen, which should result in a more homogeneous temperature distribution within the measuring length of the specimen. A direct optimization process using Ansys Workbench® resulted in “Geometry 3”. For the comparison between the experimental and simulation results, the delta-temperature (ΔT) regarding a temperature profile along the line L2 with 26 mm length was calculated (s. Figure 4).

Fig. 5
figure 5

Flux concentrator geometries for the induction heating process (dimensions in mm; G=Geometry)

Full size image

With the numerical results of the concentrator geometries, a tailored geometry (s. Figure 3(c)) of the flux concentrator was designed to obtain an inhomogeneous temperature with a linear decrease of the temperature along the tensile direction of a specimen.

2.4 Dilatometer Tests

The dilatometer tests were performed to determine the accuracy of the optical method by comparing the transformation start and finish temperatures, hardness and final microstructure of the heat treated or thermo-mechanically treated flat steel specimens. In addition, the possibility of realizing multi-stage cooling was investigated for the novel method. A DIL 805A/D+T dilatometer (TA Industries) at the University of Hannover was used to perform the dilatometer tests. For the comparison of the two methods, the AlSi-coated 22MnB5 steel specimens were prepared as described in section 2.2 and then heated up to the austenitizing temperature of 930 °C with 10 K s−1. The realized cooling curves after soaking for 300 seconds are depicted in Figure 6. The first specimen was quenched from 930 °C to 600 °C with 30 K s−1, then a further cooling to 450 °C with 4 K s−1 was applied and finally a cooling rate of 24 K s−1 was realized up to a temperature of 250 °C (s. Table II condition C1). Condition C2 represents a first cooling step from 930 °C to 550 °C with 34 K s−1 and a second cooling step from 550 °C to 250 °C with a cooling rate of 7 K s−1. For the condition C3 and C4, the same cooling steps were applied as for C2, but a non-isothermal hot deformation of 10 pct (strain rate of about 0.009 s−1) for C3 or 20 pct (strain rate of about 0.018 s−1) for C4 was performed simultaneously during the first cooling step (s. Table II).

Fig. 6
figure 6

Cooling curves for the comparison between the optical method and dilatometer tests

Full size image
Table II Cooling Steps of the Multi-stage Cooling
Full size table

For determination of the transformation start and finish temperatures of both methods, dilatometer curves are generated in Excel and tangents were applied to the dilatation curves to visualize the transformation temperatures.[6]

Additionally, the results of two phase transformation analysis using the optical method were compared with dilatometer curves presented by Naderi et al.[34] Two soaking durations were applied to reveal possible differences. Therefore, specimens with the dimensions shown in Figure 1(c) were austenitized at 900 °C, soaked for 300 or 600 seconds and then quenched by compressed air with a cooling rate of 50 K s−1. For the 600 seconds soaked specimen a tensile deformation was performed after the end of the soaking until a total elongation of 20 pct was reached.

2.5 Graded Thermo-mechanical Processing

To demonstrate the application fields for the optical method tapered tensile specimens were investigated to analyze the influence of different degrees of deformation on the transformation kinetics with only one specimen. In this paper, two experimental examples pointing out a strain gradient within one specimen and a combination of a strain and temperature gradient within one specimen will be presented in Section III–C. A targeted strain gradient was reached with a 1 deg tapering of the measuring length (s. Figure 7). To consider only the influence of the strain gradient, the tailored flux concentrator geometry with the lowest ΔT (s. results of Section III–A) was chosen to obtain a homogenous temperature distribution along the measuring length of the tapered specimen. The hot tensile test was performed, as described in Section 2.2, with the specimen being drawn isothermal after soaking to just before failure. Then, the cooling to the ambient temperature in free air was initiated. A 5 deg tapered specimen was used to demonstrate the combination of a strain and temperature gradients. A tensile deformation of up to 21 pct took place isothermally after soaking at the austenitizing temperature. Afterwards, the specimen was cooled down with compressed air to the ambient temperature. The realization of a temperature gradient along the tensile direction of a specimen by a targeted application of the flux concentrator was demonstrated in Reference 14.

Fig. 7
figure 7

Flat tensile specimen with a tapered measuring length

Full size image

The applied optical method enables the approximation to real process conditions, since also in the real process certain temperature and strain gradients occur.[35] Especially in tailored austenitizing processes, gradients in the cooling rate are generated.[36]

2.6 Microstructural and Mechanical Properties

After the thermo-mechanical treatment, metallographic samples are taken from the dilatometer specimens. The whole measuring length of the optically analyzed specimens was embedded in epoxy resin as demonstrated in the flow chart of Figure 2 to enable a complete investigation of the specimen’s microstructure. All metallographic samples were grinded, polished and then etched in 3 pct nitric acid dissolved in ethanol (Nital etching). The resulting microstructures of the areas of interest were analyzed with the scanning electron microscope (SEM) Zeiss Ultra Plus, which was operated at an acceleration voltage of 20 kV using an in-lens detector. Pictures for a quantitative determination of the microstructural compositions were taken at a magnification of ×1000 and were evaluated using surface analysis software ImageJ as shown in the flow chart of Figure 2 and demonstrated in Reference 37. Further, the determined final microstructure compositions were compared with the hardness values and the measured transformation temperatures.

An automatic hardness tester KB 30 FA was utilized to test the hardness with a force of 9.807 N (HV1). Five measurements per sample were carried out according to DIN EM ISO 6507 for the comparison between the dilatometer tests and the optical method. For the graded thermo-mechanically treated tapered specimens, a core hardness curve along the tensile direction was determined as shown in the flow chart of Figure 2.

3 Results and Discussion

3.1 Influence of the Flux Concentrator Geometry on the Temperature Distribution

The results of the simulations and experimental verifications of the three flux concentrator geometries presented in Figure 5 are depicted in Figure 8(a). For the initial state representing a constant u-shaped flux concentrator geometry (s. Geometry 1 in Figure 5), a bell shaped curve is visible with a ΔT of ±18 °C for the measuring length L2 (26 mm). The simulation of this concentrator shape demonstrates a good match of the model with a calculated ΔT of ±14 °C. With Geometry 2 milled with a radius of about 121 deg, a significant improvement of the homogeneity was reached with a ΔT of ±6 °C for the experimental verification of the simulation (ΔT= ±2 °C). The numerical simulation of the improved concentrator with Geometry 3 revealed a better temperature homogeneity with a ΔT of ±1 °C, but the experimental test of Geometry 3 did not prove a further advance in homogenization of the temperature with a ΔT of ±6.5 °C. Nevertheless, the shape of the curve from Geometry 2 revealed the minimum temperature in the middle of the line L2 in relation to the shape of the concentrator, whereas this effect was eliminated in case of the Geometry 3.

Fig. 8
figure 8

(a) Simulation results of the implemented induction heating model and experimental verification of the investigated three different flux concentrator geometries: Temperature values for the 26 mm long line L2 (b) possible adaption of the flux concentrator geometry to get a linear temperature gradient

Full size image

A possible reason for the discrepancies between the simulation and experimental results could be the deviation from parallel alignment between the specimen and the inductor in the experiment. Furthermore, an additional heat dissipation took place due to the clamping of the specimens, which was not considered in the simulation model.

3.2 Comparison Between the Optical Method and the Dilatometer Tests

The results of the phase transformation analysis for the four thermo-mechanical processing conditions shown in Figure 6 are depicted in Figure 9 and the graphically determined transformation temperatures are listed in Table III.

Fig. 9
figure 9

Dilatation curves produced by the phase transformation analysis using the optical method (OM) and the dilatometer tests (D) (a) condition C1 (b) C2 (c) C3 (d) C4

Full size image
Table III Phase Transformation Temperatures of the Optical Method (OM) and the Dilatometer Tests (D)
Full size table

For the first condition C1, with three cooling steps, a bainite start temperature (Bs) of 609 °C and a bainite finish temperature (Bf) of 422 °C was determined for the black dilatation curve regarding the dilatometric test. Further, the enlarged section reveled no martensitic transformation immediately after the bainitic transformation (s. Figure 9(a)). At least a very low oscillation would indicated a low amount of this phase fraction as mention in References 14, 34. The orange dilatation curve of the optical method revealed similar results for the Bs and Bf temperatures with 604 °C and 420 °C, but also showed a martensite start temperature (Ms) of 420 °C and a martensite finish temperature (Mf) of 304 °C. That a martensitic transformation could be measured was probably due to the non-exact reproducibility of the planned multistage cooling by the regulation equipment of the optical method as shown in Figure 10. After the beginning of the bainitic transformation a strong oscillation of the dilatation curve was visible, which resulted due to the slower regulation of the change between the cooling rate from 30 to 4 K s−1. In Figure 9(b), the dilatometer test (black curve) of condition C2 revealed a Bs of 554 °C, a Bf/Ms of 413 °C and a Mf of 281 °C for a cooling rate of 34 K s−1 to a temperature of 550 °C and further cooling with 7 K s−1 to room temperature. For the transformation temperatures determined by the optical method (orange curve) similar values are obtained with Bs of 556 °C, a Bf/Ms of 415 °C and a Mf of 291 °C. With the same cooling steps and an application of a strain rate of 0.009 s−1 during cooling from 930 °C to 550 °C, condition C3 showed only a bainitic transformation with Bs at 557 °C and Bf at 415 °C for the dilatometer test (s. Figure 9(c)). The application of a strain during cooling led to a shift of the bainitic transformation to higher temperatures and to no formation of martensite for the applied cooling steps as mention in References 28, 34, 38. Naderi et al.[34] explained this effect with the mechanical stabilization of austenite and the acceleration of diffusion controlled transformations. The same result was obtained for the optical method with 573 °C for the Bs and 417 °C for the Bf. A further increase of the strain rate up to 0.018 s−1 resulted in a further raise of the Bs temperature up to 578 °C for the dilatometer test and 581 °C for the optical method (s. Figure 9(d)). The Bf amounted to 417 °C for the dilatometer test and 410 °C for the optical method.

Fig. 10
figure 10

Targeted temperature profile of condition C4 and resulting temperature profiles due to the test regulation of the systems dilatometer and optical method

Full size image

In summary, all four measurements of the optical method are in good accordance with the results of the dilatometer tests. The maximum deviation was found for the undeformed state C2 with 3.5 pct for the martensite finish temperature (Mf). The determined bainite start (Bs), bainite finish (Bf) and martensite start temperature (Ms) using the optical method were closer to the determined values in dilatometer tests compared to the other specimens. Possible sources of deviations could be independent of the method and the accuracy could be optimized with a change in the equipment used in the experimental setup. The utilized camera of the optical method had a much lower recording rate compared to the dilatometer (2 Hz compared to 7 Hz). This mismatch can be reduced using a camera with a higher recording frequency. Additionally, small deviations could also be the result of a not exact reproduction of the targeted temperature profile, which is shown in Figure 10. The temperature control of the dilatometer allows a more accurate reproduction of the planned temperature profile (dashed line) for all 4 conditions. An optimization of the control parameters could improve the accuracy of the obtained temperature profile of the specimen C4_OM and the other specimens. As shown in Figure 9(a) the mentioned oscillation after the Bs could led to inaccurate results if a change in cooling rate occurs in the beginning or end of a phase transformation. Therefore, an accurate temperature control is essential.

SEM images for both methods after the thermo-mechanical treatment for condition C1–C4 are depicted in Figure 11. In Figure 11(a), the microstructure after the dilatometer test of condition C1 consisted predominantly of bainite with some small martensitic islands, although no martensitic transformation was seen in the dilatometer curve. Mostly upper bainite was formed as shown by Caballero et al. for a isothermal transformation of a medium carbon steel.[39] As Zajac et al.[40] notes, the typical morphology of upper bainite consists of elongated laths arranged in packages with cementite at the lath boundaries. The microstructure of the corresponding test using the optical method also showed predominantly upper bainite with some martensitic grains (s. Figure 11(b)). The resulting microstructure compositions after the graphical analysis (s. Figure 2) and hardness of both specimens for the condition C1 are summerized in Figure 12. For C1 analysed using the optical method, 7 pct martensite was detected, whereas the dilatometer test showed a lower martensite content of about 1 pct. The higher martensite content was reflected in a higher hardness of 292 HV1 compared to 274 HV1 of the dilatometer test. In Reference 15 a fully bainitic microstructure reached a hardness of 275 HV1 after a austenitizing at 950 °C for 300 seconds and cooling rate of 11 K s−1, which is in good agreement with the microstructure of the dilatometer test. Figure 11(c) shows a mixture of bainite and martensite for the dilatometer test of the condition C2 (2 cooling steps, no deformation). As indicated by the dilatometer curve of Figure 9(b), more martensite was observed. The morphology of bainite changed to lower bainite with a precipitation of cementite within the ferrite laths.[40,41] The microstructure of the condition C2 obtained by the optical method showed the same phase fractions with a lower amount of martensite (s. Figure 11(d)). Both methods revealed a difference of 4 pct between the martensite content (s. Figure 12). The difference in hardness amounted to 15 HV1. Min et al.[42] obtained a hardness of 339 HV5 with 86 pct bainite and 14 pct martensite after austenitizing at 900 °C for 300 seconds, quenching to 420 °C with 30 K s−1, isothermal hot deformation with a strain of 0.206 and further quenching with 30 K s−1 to room temperature, which is in agreement with the microstructure of the optical method with 348 HV1 for 83 pct bainite and 17 pct martensite. Figures 11(e) and (f) showed a similar microstructure with 100 pct bainite for the dilatometer test and the optical method and revealed a degenerated form of bainite as introduced by Zajac et al.[40] The fully bainitic microstructure, which was detected by the dilatometric curves for the 10 pct hot deformed condition C3 (s. Figure 9(c)), could be verified by the microstructure. The hardness of both specimens was at the same level with about 250 HV1, which was about 20 HV1 lower compared to the dilatometer test of condition C1 with an almost bainitic microstructure with an additional low amount of martensite (s. Figure 12). The microstructure of the last condition C4, with a hot deformation of 20 pct applied to the specimen, showed a fully bainitic microstructure for the dilatometer test (s. Figure 11(g)) and a bainitic matrix with small martensitic islands for the optical method (s. Figure 11(h)). The final microstructure composition for the specimen C4, obtained by the optical method, resulted in 2 pct martensite and 98 pct bainite (s. Figure 12).

Fig. 11
figure 11

SEM images of specimens (a) C1_D, (b) C1_OM, (c) C2_D, (d) C2_OM, (e) C3_D, (f) C3_OM, (g) C4_D, (h) C4_OM

Full size image
Fig. 12
figure 12

Resulting microstructure and hardness for the dilatometer vs optical method comparison

Full size image

Compared to the dilatometer test of C1_D, a higher martensite content of the specimen C4_OM could be proven due to a 8 HV1 higher hardness, although the dilatometer curve did not show a significant deviation in the martensite area (s. Figure 9(d)). The hardness of the dilatometric sample C4_D was on the same level as C3_D with about 250 HV1, which also exhibited a complete bainitic structure. This supports the statement that small amounts of martensite had to be present in sample C1_D, as the hardness of 274 HV1 was similar to sample C4_OM with 282 HV1, and more than 20 HV1 higher compared to the fully bainitic microstructures of sample C3_D and C4_D with 250 HV1.

The above-mentioned deviations of the targeted temperature curve shown in Figure 10 in case of the optical method could led to slight differences between microstructures obtained from both methods. It can be noted that for a multi-stage cooling an exact temperature control is essential to get a targeted microstructure composition.

In Figure 13 the results of the comparison between literature values of Naderi et al.[34] (blue curve) and the hot tensile tests observed by the optical method (orange curve) are displayed. In the table of Figure 13(a), the maximum deviation of the determined transformation temperatures amounted to 8 °C (2 pct) for a specimen austenitized at 900 °C for 300 seconds and quenched with a cooling rate of 50 K s−1. A second process route with a specimen austenitized at 900 °C for 600 seconds, a hot deformation of 20 pct and cooling with a rate of 50 K s−1 led also to comparable values for the martensitic transformation temperatures. The small deviations could be the result of the lower recording frequency, described at the beginning of Section III–A.

Fig. 13
figure 13

adapted from Naderi et al.[34]) (b) 20 pct hot deformed samples austenitized at 900 °C for 600 s (blue curve adapted from Naderi et al.[34])

Comparison between literature dilatometer curves34 in blue and dilatation curves determined by the optical method in orange (a) undeformed samples austenitized at 900 °C for 300 s (blue curve

Full size image

3.3 Examples of Graded Thermo-mechanical Processing

An example of a phase transformation analysis under influence of different strains within one specimen austenitized at 950 °C (soaking time: 300 seconds, cooling rate: 13 K s−1) is depicted in Figure 14. Changes between data points are illustrated with the dashed line to highlight them.

Fig. 14
figure 14

Example for a strain gradient within a tapered flat specimen austenitized at 950 °C for 300 s and cooled with 13 K/s (a) sample evaluation in GOM Correlate (b) transformation kinetics for the 3 areas of interest (c) resulting microstructure and hardness

Full size image

As seen in Figure 14(a), three different elongations are investigated, whereby the whole specimen area could be continuously analyzed with values of the elongation at hot deformation from 15 to 180 pct. Using the optical method, the transformation temperatures for an elongation at hot deformation of 20, 40, and 60 pct were determined (s. Figure 14(b)). An increase of the elongation at hot deformation led to an extended ferritic area, as shown in References 29, 38. An elevated ferrite starting temperature (Fs) of about 800 °C due to hot deformation was also observed in References 14, 38, 43. Naderi et al.[34] explained this effect with the increase of the stored energy of deformation, which led to a shift of diffusion controlled transformations to shorter transformation times and higher temperatures. Further, the pearlite start temperature (Ps) decreased from 700 °C to 640 °C with an increase of the elongation from 20 to 40 pct. At 60 pct a further reduction to 626 °C was determined. Schaper et al.[43] also observed a reduction of the Ps from 680 °C to 660 °C (austenitization at 950 °C, soaking duration 300 seconds, isothermal hot deformation at 800 °C, cooling rate of 15 K s−1) with a raise of the elongation at hot deformation from 20 to 40 pct. The Bs declined slightly and no considerable changes can be detected for the martensitic transformation temperatures (s. Figure 14(b)). In References 14, 43, the same effects for the Bs and Ms were observed after fully austenitization, hot deformation and cooling with 12 or 15 K s−1. A reduction in Ms with increasing strain as in Naderi et al.[34] was not found. On the one hand, this could be due to the relatively slow cooling, as a result of which only a small amount of martensite was formed (s. Figure 14(c)). On the other hand, due to the low recording rate of 2 Hz, the resolution might not have been sufficient to show this effect.

In case of the microstructure compositions, a decreasing bainite content and an increasing ferrite content are determined with a growing elongation at hot deformation (s. Figure 14(c)). This effect was also shown in Reference 14 as a result of an extended ferritic area for a short time austenitized specimen (900 °C for 10 seconds and cooling with 12 K s−1). After an increase of the elongation from 40 to 60 pct, a reduction in the martensite content and a slight increase of the pearlite content were observed. The reduction of martensite by raising the hot deformation strain was also showed by various authors.[28,34,38,38,44,45] As a consequence of hot deformation, a stabilization of the austenite phase region occurs, which led to a shift of the softer phases to shorter transformation times and therefore to a higher amount of these in the final microstructure for cooling rates beneath the critical cooling rate.[34,38]

The hardness values (green crosses) correlated with changes in the microstructure, where a 14 pct increase of the ferrite content due to a growing elongation from 20 to 60 pct resulted in a 5 pct lower hardness. Compared to the hardness of specimen C1_OM in Figure 12 with 292 HV1 for a microstructure of 7 pct martensite and 93 pct bainite, a 10 HV1 lower hardness was determined for the 20 pct hot deformed area in Figure 14(c) for 89 pct bainite, 5 pct Martensite, 5 pct ferrite and low amount of pearlite.

An example of the phase transformation analysis of an inhomogeneously austenitized tapered specimen is shown in Figure 15. The distribution of the elongation at hot deformation and temperature gradient are depicted in Figure 15(a). Three characteristic areas were investigated, which represents different austenitizing temperatures and values for the elongation at hot deformation. The highest temperature was reached within the widest specimen section. In Figure 15(b), the obtained dilatation curves and resulting transformation temperatures are illustrated. For the first area 1, a Fs of 790 °C, a Bs of 611 °C, a Ms of 387 °C and a Mf of 274 °C were determined. Based on the deflections of the dilation curve, most of the austenite should have been transformed to bainite or ferrite and only a small amount of martensite should have been formed, even though the cooling rate was 26 K s−1 (s. Figure 15(c)). Due to an austenite start temperature (Ac1) of 723 °C and a finish temperature of primary ferrite to austenite transformation (Ac3) of 880 °C,[28] a certain amount of non-transformed ferrite should be present after austenitization. Therefore, a high ferrite content of 80 pct was determined after austenitization at 815 °C and cooling close to the critical cooling rate for an undeformed 22MnB5 steel (s. Figure 15(c)).[46] Reitz et al.[28] stated that after an austenitizing at 800 °C for 300 seconds, hot deformation of 20 pct and high cooling with 32 K/s a ferrite content of 70 pct was formed. Since the bainite content was 10 pct and the martensite content 5 pct, it can be assumed that at least 15 pct of the initial microstructure was converted to austenite during austenitization. Furthermore, it was shown in Reference [28] that the critical cooling rate for austenitizing between the Ac1 and Ac3 temperature is higher than for complete austenitization of the steel. The hardness of 221 HV1 within area 1 (s. Figure 16(a)) confirmed the high amount of ferrite, which was determined by the image analysis of the SEM pictures (s. Figure 2). Of the bainitic fractions, predominantly upper bainite was formed, as shown in Reference 40 and seen in Figure 16(a). The 6 pct of pearlite appeared both in a degenerated form[47] and in the form of small islands.[29]

Fig. 15
figure 15

Example for an inhomogeneous treated tapered specimen with a temperature and strain gradient soaked for 300 s and quenched by compressed air (a) sample evaluation in GOM Correlate, (b) transformation kinetics for the 3 areas of interest, (c) resulting microstructure and cooling rate

Full size image
Fig. 16
figure 16

SEM images and hardness values for the inhomogeneous treated tapered specimen (a) area 1 austenitizing temperature 815 °C, (b) area 2 austenitizing temperature 845°C, (c) area 3 austenitizing temperature 860 °C

Full size image

For area 2 austenitized at 845 °C, a Fs of 786 °C, a Bs of 604 °C, a Ms of 387 °C and a Mf of 281 °C were determined according to the tangents of the orange dilatation curve (s. Figure 15(b)). Westermann et al.[29] obtained comparable results for the transformation temperatures after austenitization at 850 °C for 300 seconds, 20 pct hot deformation and cooling with 29.5 K s−1, where additionally a pearlitic transformation could be detected and a final amount of 10 pct pearlite. After a hot deformation of 19 pct and cooling with 32 K s−1 a final microstructure composition of 12 pct ferrite, 2 pct pearlite, 9 pct bainite and 77 pct martensite was determined (s. Figure 15(c)), whereas in Reference 29 the bainite content amounted to about 70 pct with a 2 K s−1 lower cooling rate and final hardness of 270 HV1. As shown in Figure 16(b) the hardness amounted to 400 HV1, which is confirmed by the high proportions of martensitic grains in the microstructure. Zhou et al.[44] stated that the critical cooling rate increases up to 40 K s−1 for an austenitizing temperature of 900 °C due to the implementation of a hot deformation of 15 pct. The small differences in strain and cooling rate compared to Reference 29 could be an indication that area 2 was relatively close to the critical cooling rate. A further slight increase of the austenitizing temperature to 860 °C and cooling rate to 34 K s−1, regarding area 3 in Figure 15(c), confirms this assumption. With similar transformation temperatures (grey curve s. Figure 15(b)) compared to area 2, the martensite content was raised up to 96 pct. This was also reflected by a stronger expansion of the dilatation curve within the martensitic temperature range and the hardness with 443 HV1 (s. Figure 16(c)). The SEM image in Figure 16(c) revealed a martensitic matrix with some bainitic grains. Further, the obtained microstructure composition indicated that almost all of the austenite was transformed to martensite and therefore the cooling rate must be close to the critical cooling rate. In addition to the low content of bainitic grains, some ferritic grains were also found, which, due to the austenitization below the Ac3 temperature, should be residual ferrite. The low amount of bainite could be formed due to an inhomogeneous carbon distribution within the austenite due to undissolved carbides as stated by Andreiev et al.[48] The pearlite content declined to 0 pct after an utilization of a higher austenitizing temperature of 860 °C before quenching. The small pearlitic islands, which were discussed in Reference 29, are not formed after quenching at an austenitizing temperature of 860 °C compared to quenching from 810 °C.

In summary, it can be seen that a different austenitization temperature, elongation at hot deformation and resulting cooling rate led to slightly different transformation temperatures, whereas the highest difference was determined for the Mf temperature. The resulting microstructure compositions for a growing austenitizing temperature shown in Figures 15(c) and 16 revealed a change from an almost ferritic-pearlitic microstructure to almost martensitic microstructure. An implementation of a temperature gradient also resulted in a small gradient of the cooling rate with a highest difference of 8 K s−1 between the lowest and highest austenitizing temperature. The hardness strongly increased with the growing martensite content and reached a maximum value of 443 HV1, which is 34 HV1 lower in comparison with a fully martensitic microstructure in press-hardened steel shown in Reference 29 and exhibiting a similar chemical composition.

4 Conclusions

In this paper, the accuracy of an optical characterization method for the determination of phase transformations is ascertained by a comparison with dilatometer tests. Furthermore, the opportunities of the optical method are highlighted by two experimental setups for a graded thermo-mechanical treatment of tapered flat specimens. For a targeted regulation of the specimen temperature to reach a homogenous or graded temperature distribution, a numerical model for the geometrical adjustment of the flux concentrator was presented. The overall results can be summarized as follows:

  1. 1.

    A comparison between the results of the optical method, and dilatometer tests as well as dilatometer curves available in the literature revealed a high accuracy of 96.5 pct, for the multi-stage cooling, and 98 pct, for the single stage cooling, using the optical method. The obtained deviations can be eliminated by an exact temperature control and a higher recording frequency.

  2. 2.

    A numerical model for the induction heating process enabled the homogenization of the temperature distribution along the measurement length of the specimen thanks to geometrical adjustments of the flux concentrator. A good match between the experimental and simulation results was determined. As a result, linear temperature gradients could be implemented within one specimen to investigate a tailored austenitizing process using the optical method.

  3. 3.

    The examples of the strain gradient or the combination of a strain and temperature gradient within one specimen revealed the benefit of the optical characterization method. A design of the graded component could be performed with a reduced number of tests using the optical method. In this context, the duration of metallographic preparation could also be decreased thanks to the reduced number of specimens.

    With the used experimental setup close-to-process conditions, as, for example, a non-linear cooling or inhomogeneous deformation, can be represented.

  4. 4.

    The optical characterization method is suitable for a big data analytics thanks to the possible extension of the areas of interest on the whole specimen length, representing large number of strain and thermal states.