Abstract
The automotive industry is under increasing pressure to: (1) reduce the weight of vehicles and (2) improve crash performance. ‘Hot stamped’ mild carbon–manganese–boron steel 22MnB5 has become common place in the body structure of the European vehicle over the past decade. The ultrahigh strength martensitic microstructure resulting from hot stamping (proof strength ∼1200 MPa and ultimate tensile strength ∼1500 MPa) enables down gauging while not compromising crash performance. However, with demands for yet higher strength in the final component so to enable further down gauging, novel grades must be developed. In this paper, development of the novel grade 38MnB5 was reported. Following hot stamping, 38MnB5 demonstrated proof strength in excess of 1400 MPa and ultimate tensile strength in excess of 2000 MPa. Owing to the immense strength, the novel grade 38MnB5 was considered to offer significant down gauging and weight reduction opportunities to the automotive industry.
Introduction
The automotive industry is under increasing legislative pressure to: (1) reduce the weight of vehicles so to improve fuel efficiency and (2) improve crash performance and associated occupant safety.
‘Hot stamping’ is the revolutionary forming process for ultrahigh strength steel automotive ‘anti-intrusive’ structural body components, such as roof pillars; door, floor and roof reinforcements, and bumper beams. 1 The ‘safety critical’ anti-intrusive structural body components are designed to form a rigid ‘safety cell’ around the passenger compartment and resist deformation during impact. 2
In the hot stamping process,1,3–11 the sheet steel blank is furnace heated to 900–950°C so as to achieve a homogenous austenitic microstructure; transferred from furnace to press by a robotic transfer system and rapidly stamped (formed) into the desired component geometry. Owing to the high temperature austenitic microstructure that is maintained throughout forming, low strength (true strength: ∼200 MPa), 12 high ductility (true strain: ∼50), 12 plastic isotropicity (r value: ∼1 and Δr value: ∼0) 13 and thus, high formability are found, enabling down gauging (weight reduction) while not compromising forming limits. Moreover, multiple components (constituting a welded assembly) can be consolidated in one single geometrically complex forming operation, reducing process time (cost), reducing weight and increasing structural strength. Immediately following forming, the component is quenched to near ambient temperature while constrained in the press. Quenching transforms the highly formable austenitic microstructure to an ultrahigh strength martensitic microstructure (proof strength: ∼1200 MPa, ultimate tensile strength: ∼1500 MPa and total elongation: ∼6·0). 14 Thus, the final component is highly resistant to deformation during impact, enabling down gauging while not compromising anti-intrusive crash performance. Moreover, constraint on the component throughout quenching prevents springback and associated geometric distortion. 4
The conventional steels for automotive hot stamping technologies have become ‘boron steels’: 15 grade 22MnB5 has become the ‘benchmark’. 16 22MnB5 is a mild carbon–manganese–boron, low alloy grade (Table 1). The low alloy content promotes relative economy, cold and hot formability, and weldability. However, the minute boron content (30–50 ppm) promotes exceptional quench hardenability and attainment of the ultrahigh strength martensitic microstructure.17–21
Chemical composition (wt-) of investigated grades (Fe balanced)
Ultrahigh strength hot stamped boron steel 22MnB5 has become common place in the body structure of the European vehicle over the past decade. 22 Volvo, Porsche and VW Group in particular have made significant investments in the hot stamped boron steel technology. Outside of Europe, Ford, GM and Honda have shown strong interest in the technology.22,26
However, 22MnB5 is now widely regarded as a ‘commodity product’ and with ever increasing demands for higher strength in the final component so to enable further down gauging, novel grades must be developed. German steel manufacturer ThyssenKrupp reports the development of a boron steel grade that exhibits ultimate tensile strength in excess of 1900 MPa, combined with total elongation of more than 4·0 following hot stamping. 23 Naderi et al. report the development of a boron steel grade that exhibits ultimate tensile strength in excess of 1800 MPa, combined with total elongation of more than 9·0 following hot stamping and post-quench tempering. 7 Further, it is reported by experts within the automotive steel industry that in order to achieve significant down gauging while not compromising anti-intrusive crash performance, ultimate tensile strength in excess of 2000 MPa would be necessary. Thus, the objective of the current work was to achieve ultimate tensile strength in excess of 2000 MPa, combined with total elongation of more than 5·0 following hot stamping. Moreover, this was to be achieved with minimal loss to the relative economy, cold and hot formability, and weldability that have helped to make 22MnB5 the conventional grade for automotive hot stamping technologies. Hence, ‘rich’ and costly alloying additions were to be minimised. Further, the objective was to be achieved from simple hot stamping heat treatment, without the addition of costly post-quench heat treatment.
Investigated grades
Investigated grades included 22MnB5 as the control grade and 38MnB5 as the novel experimental grade. Chemical composition of each grade is presented in Table 1. Also presented in Table 1 is carbon equivalent (CEN). CEN was calculated according to equation (1). 24
The chemistry of 38MnB5 was determined from review of literature and subsequent laboratory experimentation. In total, seven trial boron steel grades (including 38MnB5) were tailor designed and laboratory produced. The seven grades exhibited different carbon contents. Moreover, additions of inorganic alloying elements, including molybdenum, vanadium and nickel, were also investigated. It was concluded that 38MnB5 provided the greatest potential of meeting the objective laid-out in the section on ‘Introduction’.
It is widely known that carbon increases the strength/hardness of martensite more dramatically than any other alloying element on a weight for weight basis. 25 Carbon introduces precipitation hardening potential as carbides may be precipitated, 24 where carbides may also serve to restrict austenitic grain growth and thereby, refine the quenched microstructure, where microstructural refinement of lath martensite has been reported to increase strength without decreasing impact toughness. 26 Carbon increases quench hardenability, 25 where it has been suggested that quench hardening with a cooling rate somewhat greater than the critical achieves a finer martensitic microstructure. 26 Moreover, greater quench hardenability intrinsic to the chemistry may allow for a complete transformation to martensite without such a great requirement for austenitic grain growth, also giving rise to a finer martensitic microstructure. Carbon lowers the Ms temperature, while the Ar1 temperature remains unchanged, 25 which can provide a greater (metastable) austenitic hot stamping window. Furthermore, it has been reported that with more than 0·2 wt- carbon, ‘autotempering’ of martensite becomes significantly more probable during quenching (owing to increased carbon saturation and the lower Mf temperature), which can provide enhanced elongation and impact toughness while maintaining strength. 26
Since carbon content was significantly below 0·6 wt-, martensite was expected to be predominantly of the lath character rather than the plate character. 25 Lath martensite is preferential since the substructures constituted by packets, blocks and laths provide strength and impact toughness in a microstructural refinement mode. Additionally, since carbon content was close to 0·35 wt-, the inorganic alloying elements dissolved in martensitic supersaturated solid solution were expected to contribute significantly to martensitic hardness, where it has been reported that with more than ∼0·35 wt- carbon, the contribution to martensitic hardness by the dissolved inorganic alloying elements is deteriorated. 25
It is also worthy to note that carbon represents one of the most economical alloying elements. Further, maintaining the base chemical composition approximately equal to that of the control grade 22MnB5, except for modification to the content of just one alloying element (carbon), is attractive from the perspective of industrial logistics should the experimental grade reach commercial production.
Both the control grade 22MnB5 and the experimental grade 38MnB5 were laboratory produced by Tata Steel Research Development & Technology UK. The production process involved vacuum batch casting of ingots weighing approximately 30 kg and with approximate geometry of 550×110×38 mm. The cast ingots were subsequently cut into blocks with approximate geometry of 55×110×38 mm. The blocks were reheated to 1250°C and soaked for 3 h. Following the soak time, the blocks were hot rolled to sheet with thickness of 2 mm via seven passes through a single stand two high reversing mill. Target finishing temperature was 850°C. The hot rolled sheet was cooled to a simulated ‘coiling’ temperature with target of 630°C on the runout table and with line speed of 160 mm s−1. The sheet at ∼630°C was placed in the furnace set at 630°C, the furnace was switched off and both furnace and sheet were permitted to naturally cool to ambient temperature. This final stage was to simulate the slow cooling of a coil in industrial production.
The hot band with thickness of 2 mm was then cold reduced by 25 to exhibit cold rolled thickness of 1·5 mm. Cold rolling was achieved via several passes through a single stand two high reversing mill
24
Experimental
Laboratory hot stamping
The blank with geometry of 225×25×1·5 mm was furnace heated (in unprotected atmosphere) to the desired soak temperature. Soak temperatures included 800, 850 and 900°C. Soak times included 1, 3 and 5 min. Thus, nine soak time–temperature conditions were investigated.
Heating rate depended on grade and most notably, soak (furnace) temperature. For 22MnB5, the heating rate ranged from 4·1 to 10·9°C s−1. For 38MnB5, the heating rate ranged from 4·0 to 6·7°C s−1. Heating rates were determined by a preliminary investigation, wherein thermocouple data logging was performed on blanks with geometry equal to that mentioned above. Figure 1 presents the time–temperature ‘heating curves’ for 22MnB5 and 38MnB5 respectively.
a 22MnB5; b 38MnB5
Following the soak time, the blank was manually transferred from furnace to press in ∼8 s. In the press, the blank was hot stamped and simultaneously die quenched to ∼150°C. The mean cooling rate above 500°C was in excess of −100°C s−1. The mean cooling rate between 500 and 200°C was in excess of −60°C s−1. Cooling rates were determined by a preliminary investigation, wherein thermocouple data logging was performed on the die/blank interface. Figure 2 presents the time–temperature ‘cooling curve’ for 22MnB5 soaked at 900°C. Note that accurate temperature measurement was possible from only ∼500°C downwards, due to the time required to establish full thermal contact between the thermocouple (positioned in the die) and the blank. During this time, the blank was already cooling at a rapid rate.
22MnB5 time–temperature cooling curve from 900°C soak temperature
No ‘forming’ took place during hot stamping. Rather, the blank was simply die quenched. This was principally due to geometric constraints of the tooling, as illustrated by Fig. 3. However, with the intention of taking a tensile specimen from the base of the hot stamped blank, the employed procedure was considered to be more than sufficient.
Laboratory hot stamping tool and blank geometry
Following hot stamping/die quenching, the hot stamped blank was removed from the press and naturally air cooled to ambient temperature.
Figure 4 illustrates the laboratory hot stamping heat treatment cycles.
Laboratory hot stamping heat treatment cycles
Tensile testing
The hot stamped blank with geometry of 225×25×1·5 mm was cut into a 170 mm tensile blank by removing 55 mm from the end that was handled during hot stamping. This meant that the entirety of the tensile blank received ‘homogeneous’ heat treatment. Finally, the tensile blank was machined to a standard 50 mm gauge length tensile specimen. 27 The configuration is illustrated by Fig. 5. Tensile testing was performed with the Zwick 1474 100 kN electromechanical tensile testing machine according to standard procedures. 27
Configuration of tensile specimen from hot stamped blank
Microstructural analysis and hardness testing
The micro specimen (∼11 mm in length) was taken from the widest region of the tensile specimen, which was gripped by the tensile machine during the tensile test. The end of the tensile specimen selected was that furthest from the fracture point, so as to ensure that the microstructure represented the as hot stamped condition, without any modification from the tensile test. The cross-section longitudinal to the tensile specimen was examined. The configuration is illustrated by Fig. 6.
Configuration of microspecimen from tensile specimen
Nine light optical microscopy (LOM) images and three scanning electron microscopy (SEM) images were taken per examination surface. The configuration is illustrated by Fig. 7. LOM was performed with the Reichert Jung MeF3 Light Optical Microscope. SEM was performed with the JEOL JSM 6100 SEM.
Configuration of microscopy and hardness testing points
LOM images were used for preliminary detection of microconstituents. SEM images were used for microconstituent volume fraction measurement and proeutectoid α-ferrite grain size measurement (where applicable). These microstructural analyses were performed using Adobe Photoshop software. The method involved rendering a ‘map’ of the micrograph by manually tracing around each individual proeutectoid α-ferrite grain. The proeutectoid α-ferrite grains were coloured pink, while the remaining microstructure constituted by martensite was coloured black. Microconstituent volume fractions and mean proeutectoid α-ferrite grain size could then be calculated from the pixel count. Although it is acknowledged that a published standard procedure does not exist for the above microstructural analyses using Adobe Photoshop, the method represents an ‘in house’ standard procedure that is accepted throughout Tata Steel, since it is considered to be more accurate and precise than published standard procedures28,29 that are available.
It is widely reported that quench hardened martensitic grades exhibit a volume fraction of retained γ-austenite. Moreover, the probability of retained γ-austenite is raised with increasing carbon content or carbon equivalent, since the Mf temperature is depressed. However, the retained γ-austenite usually presents itself with such a low volume fraction and with such a thin ‘film-like’ or ‘interlath’30,31 morphology that it is seldom observable from LOM or even SEM. Thus, to assess the presence of retained γ-austenite, X-ray diffraction (XRD) analysis was performed with the Philips PW1830 generator and PW1710 diffractometer controller operating at 50 kV and 40 mA using molybdenum irradiation, according to standard procedures. 32
Ten Vickers hardness measurements were taken at 1 mm intervals along the examination surface length with the Leco Micro Hardness testing machine according to standard procedures. 33 This was performed at ¼, ½ and ¾ the specimen thickness. The configuration is illustrated by Fig. 7.
Results and discussion
An overview of results is presented in Tables 2 and 3. Figure 8 presents retained γ-austenite volume fractions determined by XRD analysis.
Retained γ</emph>-austenite volume fractions determined by XRD analysis
22MnB5 results (tensile properties mean values from three samples)
38MnB5 results (tensile properties mean values from three samples)
Following a given soak condition, 38MnB5 consistently exhibited equal or higher retained γ-austenite volume fraction than 22MnB5 (except for 5 min-850°C). This can be attributed to the higher carbon content of 38MnB5 lowering the Ms and Mf temperatures. However, for both grades following all soak conditions, the retained γ-austenite volume fraction was marginal, reaching a maximum of just 3·6 vol.- for 22MnB5 and 4·5 vol.- for 38MnB5. Owing to the marginal presence of retained γ-austenite, further discussions are simplified by ignoring the presence of retained γ-austenite and are focused on the dominant microconstituents, namely, proeutectoid α-ferrite and martensite.
22MnB5 presented a ‘dual phase’ ferritic–martensitic microstructure following the soak temperature of 800°C. Increasing soak time from 1 to 3 and to 5 min at the soak temperature of 800°C resulted in higher martensite volume fraction. The completely martensitic microstructure was achieved given the soak temperatures of 850 and 900°C (Fig. 9). The above observations illustrate the correlation between soak time–temperature, austenitic grain growth and quench hardenability. Increasing soak time–temperature results in greater austenitic grain growth, smaller austenitic grain boundary surface area, fewer heterogeneous nucleation sites for reconstructive proeutectoid α-ferrite formation and thus, greater quench hardenability, giving rise to higher martensite volume fraction.34–36
a 5 min-800°C; b 5 min-850°C; c 5 min-900°C
By comparison, 38MnB5 achieved a completely martensitic microstructure given the soak condition of 5 min-800°C (Fig. 10). This illustrates the positive effect of carbon content on quench hardenability. Greater quench hardenability intrinsic to the chemistry of 38MnB5 allowed for a complete transformation to martensite without such a great requirement for austenitic grain growth.
a 5 min-800°C; b 5 min-850°C; c 5 min-900°C
22MnB5 achieved maximum proof strength Rp0·2, ultimate tensile strength Rm and hardness HV10 given the soak condition of 1 min-850°C. Given the soak temperature of 800°C, the above properties were compromised by the presence of proeutectoid α-ferrite. Given a soak condition above 1 min-850°C, it can be suggested that the above properties were compromised by greater coarseness of the martensitic microstructure (larger packet size), resulting from excessive austenitic grain growth. Conversely, the presence of proeutectoid α-ferrite resulting from the soak temperature of 800°C gave rise to higher uniform elongation Au and total elongation A50. It can be suggested that greater coarseness of the martensitic microstructure, in addition to greater microstructural homogeneity (indicated by lower standard deviation of hardness across the specimen) resulting from a soak condition above 1 min-850°C, also gave rise to higher Au and particularly higher A50. Further, it has been suggested 37 that ferrous and alloy carbides (previously found to be more abundantly nucleated with a soak condition above 1 min-850°C) may scavenge free hydrogen molecules from the primary microconstituent(s) (martensite), giving rise to higher impact toughness and elongation.
38MnB5 achieved maximum Rm given the soak condition of 5 min-850°C. Given a soak condition below 5 min-850°C, Rm was compromised by proeutectoid α-ferrite presence and/or excessive microstructural heterogeneity. The latter suggestion can be supported from inspection of the hardness profiles given the soak conditions of 5 min-800°C, 5 min-850°C and 5 min-900°C (Fig. 11): much greater homogeneity both along the specimen length and through the specimen thickness can be seen to result from the increase in soak temperature. Moreover, the soak condition of 3 min-850°C resulted in higher Rp0·2 and HV10 than the soak condition of 5 min-850°C. All soak conditions below 5 min-850°C (except for 1 min-800°C) resulted in higher HV10 than the soak condition of 5 min-850°C. Given a soak condition below 5 min-850°C, Au and A50 were very low. Indeed, for a given soak condition below 5 min-850°C, the recorded Au value equalled the recorded A50 value. This is a strong indication of premature brittle fracture, since the specimen fractured at the ultimate tensile strength (or even before the theoretical ultimate tensile strength) rather than persisting to a characteristic failure strength. Thus, it is suggested that the finer martensitic microstructure resulting from a soak condition below 5 min-850°C provided potential for maximum Rm (indicated by higher Rp0·2 and/or HV10). However, excessive microstructural heterogeneity (indicated by much greater heterogeneity of hardness, much lower Au and A50 and with the Au value equalling the A50 value) resulted in premature brittle fracture that prevented the maximum Rm potential from materialising.
a 5 min-800°C; b 5 min-850°C; c 5 min-900°C
Figure 12 presents Rm against HV10 for each soak condition. As a general rule, Rm is reported 38 to be approximately 3·0 times the HV10. Given a soak condition equal to and above 5 min-850°C, the approximation was consistent. However, given a soak condition below 5 min-850°C, the approximation was inconsistent, with Rm approximately just 2·2–2·6 times the HV10 (Table 3). It is also worthy to note the change of fracture mode (Fig. 13). Given all soak times at the soak temperature of 800°C, the fracture surface was at 90° to the gauge length, indicating a brittle pure shear fracture. 39 With increasing soak time at the soak temperature of 850°C, the fracture surface gradually became closer to 45° to the gauge length, indicating a more ductile plane nominal shear fracture. 39 Given all soak times at the soak temperature of 900°C (except for 5 min-900°C, which should be considered an anomaly), the fracture surface was at 45° to the gauge length.
Ultimate tensile strength–hardness
38MnB5 fractured tensile specimens illustrating change of fracture mode
The optimal soak condition for 22MnB5 was considered 1 min-850°C. The optimal soak condition for 38MnB5 was considered 5 min-850°C. The optimal soak condition for each grade was determined principally by maximum Rm, with maximum Rp0·2 and adequate A50 also important considerations.
With both grades treated to their optimal soak condition, 38MnB5 demonstrated significantly higher Rp0·2 (+256 MPa) and Rm (+476 MPa), yet with only a small loss to A50 (−1). The highlighted properties of 38MnB5 demonstrate superior anti-intrusive crash performance.
Conclusions
Laboratory hot stamping was performed with the conventional boron steel grade for automotive hot stamping technologies 22MnB5, in addition to the novel experimental grade 38MnB5. Nine soak conditions were investigated for each grade. The optimal soak condition for 22MnB5 was found to be 1 min-850°C. The optimal soak condition for 38MnB5 was found to be 5 min-850°C. Although the richer carbon content and hence, greater quench hardenability intrinsic to the chemistry of 38MnB5 provided a completely martensitic microstructure given a lower soak condition, the longer soak time of 5 min rather than 1 min was necessary for 38MnB5 in order to achieve microstructural homogeneity given the richer carbon content.
Following laboratory hot stamping treated to its optimal soak condition, 38MnB5 demonstrated Rp0·2 of 1419 MPa and Rm of 2066 MPa. These tensile strength properties were significantly higher than those of 22MnB5 treated to its optimal soak condition. Moreover, elongation of 38MnB5 was found to be comparable to that of 22MnB5, with A50 of 6·2.
Owing to the immense strength of 38MnB5, significantly above that of 22MnB5 and moreover, with Rm above the target of 2000 MPa, 38MnB5 was considered to offer significant down gauging and weight reduction opportunities to the automotive industry. Moreover, this was achieved with a chemistry that can still be considered relatively lean and from simple hot stamping heat treatment. However, it should be acknowledged that the highlighted benefits of 38MnB5 are purely from an academic perspective. For the highlighted benefits to be translated into industry, manufacturing constraints must be overcome, namely, cold rolling to a thickness below 1·5 mm and achievement of customer ‘as delivered’ maximum strength specifications. Both of these manufacturing constraints are deteriorated by the higher strength intrinsic to 38MnB5. Moreover, weldability of the final hot stamped component is also worthy of consideration.