Adiabatic heating and energy absorption capability of an advanced high strength steel during drop weight impact testing

Spain Abstract The manuscript focuses on the energy absorption capability of an advanced high strength steel during drop weight impact testing. The in-situ measurements of temperature during impact resistance testing of samples revealed intensive adiabatic heating effect with the peak temperature of 225 ℃ at the top of the dome resulting in the local softening effect. The drop weight impact resistance of the material under present testing condition is 90 J. Microstructural characterization demonstrated that dislocation glide and formation of substructure is the main deformation mechanism. Analysis of fracture surface of cracked samples (tested with the impact energy of > 90 J) revealed ductile failure mode. The energy required for formation of fracture surface was calculated by quantitative analysis of the 3D digital models generated using microscopy images of the fracture surface taken from different angles. This energy is negligibly lower compared to the total energy spent for plastic deformation of the


Advanced high strength steels
Nowadays, decreasing the weight of vehicles (i.e. car body light-weighting) is one of the priorities for automobile manufacturers due to the imposed safety and carbon emission regulations and limited fossil energy [1]. However, another car performance that should definitely not be weakened during light-weighting is passengers' safety, which is also a fracture behavior under different stress states/testing conditions. In the first category, various DP steel grades satisfying a broad scope of engineering requirements have been developed [8,11,12,[17][18][19]. On the other hand, the features of their mechanical behavior (such as distribution of stresses/strains between microconstituents) during plastic deformation, work hardening, shear banding, local crack initiation and growth, etc. have also been thoroughly investigated on both macro-and micro-scale, via experiments and simulation [20][21][22][23][24][25][26][27]. Impact resistance is one of the most important properties of AHSS for automotive applications. For DP steels, their impact resistance have been studied mainly by means of Charpy impact testing [13,16,[28][29][30][31][32][33][34][35][36][37], which is a standardized method to evaluate material impact toughness. An overview of Charpy impact energy values at room temperature reported for different DP steels is presented in Table 1. As can be seen, the measured values vary dramatically in the range of 1.7 to 235 J. It was shown that Charpy impact toughness of Vanadium-strengthened DP steel can be improved via replacement of pearlite by more ductile martensite [16]. Asim Bag et al. found that the Charpy V-notched impact energy of a Fe-0.16C-1.32Mn-0.44Si (wt. %) DP steel is maximum at 60% of martensite volume fraction [13]. Simulation conducted by Y. Prawoto et al. using Johnson-Cook model led to the similar optimum martensite volume fraction for enhanced Charpy impact resistance [28]. Computational modelling confirmed that both the constituent fraction and its morphology have a great influence on the Charpy impact resistance of martensite-ferrite DP steel [28]. Detailed investigation on DP 590 steel revealed that the ductile-brittle transition temperature (DBTT) is -95 ℃, which is far below general automobile service temperature [31]. Another study on a hot-rolled DP 590 steel grade concluded that: 1) the Charpy impact energy in ductile-brittle transition range can be increased by splitting caused by silicate and carbide inclusions; 2) splitting reduces DBTT as well [36,37]. The survey on AISI 4340 steel with different microstructure composition via adjusting heat treatment demonstrated that the combination of bainite and martensite has better Charpy impact performance and tensile ductility than martensite-ferrite or full bainite microstructure [33]. The study on AISI 3115 steel pointed out that the impact strength of this steel decreases with the increasing intercritical annealing temperature, which affects the microstructure significantly [34]. It should be noted that, bars with dimensions of 10×10×55 mm are used for Charpy impact testing [38,39], so this method cannot be utilized for characterization of common automobile AHSS steel sheets. Moreover, Charpy test is performed on notched samples, so deformation process is not relevant to the scenario, where automobile components are confronted during crash accidents. Fe-0.07C-1.52Mn-0.34Si-0.05Nb-0.05V 25 10×10 235 [42] Note: Information on the microstructure (phase volume fraction, grain size, morphology, etc.) is not provided in the table.
There have also been studies focused on impact performance of thin-walled [43][44][45][46][47] and flat sheet [48,49] specimens. Thin-walled columns composed of various cross-section shape and having various size were examined due to their similarity with automobile mid-rails.
Experimental crush testing of hydroformed thin-wall steel tubes combined with simulation exhibited that DP steels had higher energy absorption capability compared with the high strength low alloy steel (HSLA) 350 and deep drawing quality steel [45]. Modeling and experiments on DP 800 steel manifested that the axial crushing performance of thin-walled specimen has great relationship with the shape of the sample (top-hat or square section) [44].
The study on DP steels with tensile strengths from 270 MPa to 1470 MPa by K. Sato et al. showed that yield strength is the most influencing factor on the bending moment in bending crash [50]. James R. Fekete et al. estimated that energy absorption capability of thin-walled samples increases by 10% when DP 340 is substituted by HSLA 340 steel [46]. technique revealed that the plates' resistance against perforation is closely connected with the shape of the impactor (e.g. blunt-ended or ogival-ended) [48]. Another successive study completed by that team on single DP 600 sheet demonstrated significant effect of strain rate and steel grade on the strain distribution during impact deformation [49].
From this literature overview, a few gaps exist in the area of energy absorption capability of advanced high strength steels. No systematic studies on impact resistance of flat sheets tested in the conditions relevant to crash accidents have been carried out up-to-date. The research done in this field has 'engineering' character, and no much attention has been paid to microstructure evolution and adiabatic heating during dynamic plastic deformation, despite their significant role. Therefore, the main objective of the present work is to study the drop weight impact resistance of an advanced high strength steel with respect to the microstructure evolution and adiabatic heating during dynamic plastic deformation.

Material
The material studied in this work is a commercial advanced high strength zinc galvanized DP 1180 steel. The chemical composition of the steel is shown in Table 2. As a basic alloying element in ferrous alloys, 0.12 wt.% carbon was added mainly for martensite strengthening and avoiding significant reduction in weldability [2]. Manganese is highly beneficial for hardenability and austenite stabilization. To prevent formation of banded microstructure, its content was limited to 2.48 wt. %. Chromium also stabilizes austenite and increases hardenability, but its content was confined to 0.59 wt. % because of cost control. Microalloying by 0.023 wt. % of niobium was done for grain refinement effect. The material was supplied in form of 1 mm thick sheets.

Mechanical property characterization
Dog-bone tensile samples with a gauge length of 25 mm according to ASTM standard [51] were machined via electric spark cutting along the rolling and transverse direction. Their geometry is shown in Fig. 1. Tensile tests were carried out using a universal electromechanical testing machine (Instron 3384) with a constant crosshead speed of 1.5 mm/min, which corresponds to the initial strain rate of ~10 -3 s -1 . where ε is the true strain, ho is the initial thickness of the sheet and hf is the thickness at the top of the dome after testing, as schematically shown in Fig. 3. Fig. 3. Schematic presentation of a cross-sectioned sample [53]. The area marked by white square refers to the region selected for EBSD and TEM characterization.

Microstructure characterization
To reveal the microstructure of the as-received material, samples with mirror-like surface were prepared following standard metallography grinding and polishing procedures using 0.25 μm paste at the final stage and etched using 12 % Na2S2O5 water solution. A systematic manual point counting method [54]

Fracture surface reconstruction
A detailed quantitative analysis of fracture surfaces formed after cracking of samples tested with impact energy of > 90 J was performed by employing a SEM-based microscopic surface reconstruction technique [56][57][58]. This microscopic topography technique can be briefly described as following: a pair of SEM images, which were eucentricly tilted to different angles (0-10°), were used to create a 3D digital elevation model (DEM) by an algorithm which quantifies the surface height variation via searching and comparing the homologous points in the image pair [58]. Thus, one more dimension (depth, z axis) is granted to the two dimensional image, providing the possibility to measure dimple depth. In present study, SEM image pairs consisting of 1536×1103 pixels were taken at 0° and 5° tilting. A commercial MeX software (Alicona, Graz, Austria) was used to reconstruct the DEMs of fracture surface and its further quantitative analysis. The fracture surface profiles were extracted from the obtained DEMs and the depth of at least 40 dimples was measured. The outcomes of these measurements were used for estimation of the energy consumed for formation of fracture surface during impact.

Microstructure and tensile mechanical properties of the as-received material
The microstructure of the as-received material is shown in Fig. 4a. Martensite and ferrite are main microstructural constituents. Martensite has lathy and blocky shape (e.g., blue and yellow arrows in Fig. 4b), while ferrite has quasi-equiaxial grains (selectively marked by red arrows in Fig. 4a). The volume fraction of ferrite is 25.5±4.5%. The spatial distribution of both martensite and ferrite is random, and the grain size of ferrite is in the range of ~1 µm to ~5 µm ( Fig. 4a and 4b). There is also very low amount of ultra-fine grained austenite (0.23% determined by EBSD), as marked in green color in Fig. 4b. It should also be noted that the as-   Table 3.   An adiabatic thermodynamic environment can be presumed during drop weight impact tests because of the very short deformation time (<2 ms). Fig. 6c illustrates the outcomes of insitu temperature measurements drop weight impact testing with the impact energy of 90 J. From the curve, it is seen that the temperature at the top of the dome surges to the peak value of ~180 o C within ~50 ms followed by its steady decrease down to room temperature within ~5 s.
However, it should be noted that the deformation time in the experiment does not exceed 1.5 ms, as shown on the load-impact time (Fig. 6e) curve from the given experiments. This time lag can be ascribed to (1) heat transfer from the DP steel to the welded thermocouples, and (2) a thermal inertia of thermocouple itself [59]. The peak temperature tends to increase with Submitted to Materials Science and Engineering A, November, 2019 increasing impact energy (Fig. 6d) due to higher amount of plastic deformation induced into sample (Fig. 6b).

Microstructure evolution during drop weight impact testing
A thoroughly microstructural analysis of the material before and after testing was performed. Fig. 7 illustrates typical KAM maps of the material before and after impact test with 90 J energy. In the as-received material (Fig. 7a) is very low (9.5%). Impact testing with 90 J significantly increases the local misorientations all over the microstructure (Fig. 7b) and the fraction of non-indexed pixels (to 48.7%), indicating significant accumulation of lattice defects. The statistic KAM distribution is presented in Fig. 7c, evidently showing the shift of the curve from lower to higher KAM angle.  consists of equiaxial grains with diameter of up to ~6 µm (Fig. 8a). Biaxial stretching combined with plastic bending during impact testing results in flaser-like grains with aspect ratio of 0.46.
Orientation gradients in the interior of individual elongated grains can be noticed, as shown in the magnified IPF map in Fig. 8c. The detailed misorientation along the black lines in Fig. 8c are plotted in Fig. 8d, where low angle grain boundaries can be seen directly, demonstrating the formation of substructures in the grain interior. The microstructure evolution during impact loading under biaxial stress was investigated in more detail by means of TEM. Typical TEM images of the as-received material are shown in Fig. 9. The martensite laths having a width of 100-300 nm, and ferrite, which has blocky shape, are clearly seen (Fig. 9a). Microalloying by Nb also results in formation of nanoscale spherical NbC precipitates having a size of 10-20 nm, though their volume fraction is very low (Fig. 9b). Their presence was confirmed by analysis of CBED patterns, as shown in inset in Fig. 9b, where a spot corresponding to the NbC precipitate is marked by red circle. Local EDS analysis also detected Nb segregations on those nanoprecipitates (Fig. 9c). The sample after impact testing shows markedly different microstructure. It is characterized by presence of dislocation tangles and irregular dislocation cell structure in the interior of individual grains (Fig. 9d). Size of cells is in the range of 0.3-1.0 µm, which is in a good accordance with the outcomes of the EBSD analysis ( Fig. 8c and 8d). As reported previously, cell substructure can form at only 2% of true plastic strain in DP steel under uniaxial stress mode [60]. And the strain distribution inside the martensite-ferrite microstructure is normally inhomogeneous because of the higher hardness of martensite compared to ferrite. As in the present study, the accumulated plastic strain reaches up to 81.1% in the 90 J impacted specimen (Fig. 6b), well pronounced cell substructure is formed at such high plastic strain. Its formation can be related entirely to high strain rate deformation upon impact, while the effect of adiabatic heating on the microstructure can be ruled out due to low maximum homologue temperature (which reaches just 182 °C as seen in Fig. 6d) and very short time at the peak temperature (which does not exceed 50 ms, Fig. 6c and 6e). The increased KAM angle in the impacted specimen suggested the much more accumulated local deformation, mainly derived from generation of dislocation multiplication as dislocation glide is the major deformation mechanism.

Fracture surface analysis
To understand failure of the DP steel under impact loading, fracture surface of cracks formed after testing with >90 J energy was carefully examined. Fig. 10a shows typical SEM images of 95 J impacted specimen. It is seen that the cracks show ductile fracture surface with well pronounced dimples (Fig. 10a). Dimples with size varying in the range of 1-6 µm are homogeneously distributed over the fracture surface, suggesting the failure process of void nucleation, growth, and coalescence. It is not possible to distinguish the origins (martensite or ferrite) of the dimples because they are homogenous from an overall view despite the local difference. This means there is excellent deformation compatibility between hard martensite and soft ferrite, and both of them served as nucleation cites under quasi-biaxial stress mode.
The hardness difference between martensite and ferrite should be compensated by the abundant dislocations multiplied in the deformation process (Fig. 9d), as the increased dislocation density strengthened the ferrite phase. Some of the very coarse dimples were formed at manganese sulfide (MnS) inclusions were also found to be as the huge dimple initiation sites, leading flat cleavage surfaces, as shown in Fig. 10b. As this is a commercial steel and MnS inclusions are quite difficult and costly to be completely removed when the size is below 20 µm [61], the appearance of slight MnS inclusions with a size of a few micrometers is not surprising. Their influence on deformation process and energy absorption capability of the studied DP steel will not be further discussed because of their negligible fraction.
To estimate the energy consumed during fracture process, the dimple depth was quantitatively analyzed. Fig. 10c presents the reconstructed 3D DEM of the fracture surface in The depth of dimples was statistically counted from the reconstructed DEMs. For each dimple considered, its deepest elevation value was measured even though the depth inside every dimple varies from one site to another. Dimples with diameter > 4 µm should absorb significantly more energy than finer ones during cracking process. Over 40 dimples with size > 4 µm were analyzed. The averaged depth obtained for this DP steel is 1.33 µm. According to the Stüwe model [62,63], the specific energy necessary to form a unit micro-fracture surface, Rsurf can be described using the following equation after approximation: Where S=0.25 was adopted as the fracture surface is mainly consisted of dimples [62]. ℎ is the average dimple depth and ̅ is the mean flow stress of material, described as [64]:

Discussion
The local stress state on the specimen is changing continuously during drop weight impact tests, resulting in multi-axial stress state (tension, shearing, bending, etc.) in most area on the hemisphere sample. For the sake of simplification, the top of the dome-shape sample was assumed to have quasi-biaxial stress state. In addition, the energy spent for possible slight friction was also neglected.

Deformation mechanisms during drop weight impact tests
Analysis of experimental results shows a dramatic difference in ductility of the material tested in uniaxial tensile mode (8% of true strain, Fig. 5) and quasi-biaxial mode (81% of true strain, Fig. 6b). This observation can be rationalized based on two effects. First, the DP steel was softened by the abundant heat converted from the mechanical work. As the drop weight impact test was completed in only a few milliseconds and the heat generated by deformation had no time to dissipate into environment, the whole testing process can be considered as adiabatic. The adiabatic heating at the top dome-shaped sample reaches 182 ℃, which is sufficient to weaken the tensile strength of the steel. According to the previous study, the tensile strength of the DP600 steel at 200 ℃ decreases by 10-15% compared to the room temperature [65]. Similar results can be found for other steel grades in [66,67]. Therefore, the softened DP steel showed better quasi-biaxial stretching formability. Second, dislocation multiplication reinforced the ductility of the DP steel. After deformation under equi-biaxial stress state, the c d equiaxial grains were 'compressed' into 'flattened' ones (Fig. 8a, 8b and 8c), demonstrating the improved plasticity. As dislocation glide is the only deformation mechanism in the DP steel derived from its martensite and ferrite microconstituents, the 'flattened' grains should be the result of dislocation movement and multiplication of different slip systems. On the other hand, the softened material and increased temperature, both are caused by adiabatic heating, are two promotors for dislocation nucleation and motion [68].

Energy absorption capability
The work done by the drop weight mass is transformed into two parts: (1) heat, which caused temperature changing of the material and then dissipated into the environment; (2) strain energy, which was mostly stored in the material in the form of lattice defects. For the first part, it was detected directly by the thermocouples, as shown in Fig. 6c and 6d, obviously manifesting the fact that there is a lot of energy has converted into heat. The exact ratio of mechanical work transformed into heat (designated as β) has been reported previously in detail in [69][70][71], where Kolsky pressure tensile bar and infrared camera were used to satisfy an adiabatic condition and to catch the temperature rising, respectively. The ratio value measured experimentally is usually in the range of 0.8-0.9, which means most of the mechanical work during plastic deformation (i.e. 72-81 J) is spent for material heating. Consequently, the strain energy stored during plastic deformation, i.e. the second part, consumes a small portion of the total work (i.e. 9-18 J). However, the strain energy, determined by the plastic strain which limited by the ductility of the material, is mutually influenced with the total plastic work.
Because improved plasticity increases total strain as well as total mechanical work. And the total work finished before failure during drop weight impact tests equals to the energy absorbed by the tested material. Of course, the work is also influenced by the strength of the material.
For the studied DP steel, the tensile strength is above 1200 MPa, which is excellent.
Nevertheless, the ductility is not so attractive. However, this deficiency was enhanced to a large extent under dynamic biaxial loading, resulting in improved formability, as well as total plastic work, namely overall absorbed energy.

Conclusions
The microstructure evolution and energy absorption capability of a commercial high strength martensite-ferrite dual phase (DP) steel during drop weight impact testing were thoroughly investigated. The main conclusions are: 1. Intensive adiabatic heating effect was detected via in situ temperature measurements resulting in the peak temperature of 225 ℃. The drop weight impact resistance of the DP steel under present testing conditions is 90 J. The ductility of the material was improved by the softening effect due to the large amount of heat converted from plastic work.
2. Dislocation glide and formation of dislocation cell substructure is the main deformation mechanism active in the DP steel during high strain rate quasi-biaxial stretching.
3. Ductile failure mode features the fracture surface of the DP steel under high strain rate biaxial stress loading and the energy consumed during cracking presents very low fraction of the total plastic work.