You are on page 1of 9

Materials Science & Engineering A 820 (2021) 141518

Contents lists available at ScienceDirect

Materials Science & Engineering A


journal homepage: www.elsevier.com/locate/msea

Toughness enhancing mechanisms in age hardened Fe–Mn–Al–C steels


Nathan A. Ley a, Marcus L. Young a, *, B. Chad Hornbuckle b, Daniel M. Field b, Krista R. Limmer b
a
University of North Texas, Department of Materials Science and Engineering, Denton, TX, 76203, USA
b
U.S. Army Combat Capabilities Development Command Army Research Laboratory, Weapons and Materials Research Directorate, Aberdeen Proving Ground, MD,
21005, USA

A B S T R A C T

Continued research on advanced high strength steels is an important step in reducing weight and fuel consumption of modern vehicles. Fe–Mn–Al–C steels that have
increased concentrations of both Al and Mn, are currently under development to achieve these goals. Alloys with sufficiently high Al and C contents are precipitation
hardenable, strengthening the steel by forming nano-sized κ-carbide precipitates, (Fe,Mn)3AlC. In this study, the toughness and phase stability of a Fe–Mn–Al–C steel
was investigated by examining evolution of the γ-austenite matrix and κ-carbide precipitates using scanning electron microscopy (SEM), transmission election
microscopy (TEM), and synchrotron radiation x-ray diffraction (SR-XRD). Impact toughness was examined using hardness and Charpy impact testing, and results are
compared with CALPHAD predictions. This study illuminates proposed mechanisms for simultaneously improving strength and toughness in age-hardenable
Fe–Mn–Al–C steels by examining the effect of various heat treatments on the γ-austenite matrix and κ-carbide precipitates.

1. Introduction nano-sized κ-carbide precipitate [1,2,8–13]. After age hardening, these


steels have been shown to have significantly increased yield strengths
A significant body of work has been produced on advanced high with toughness equivalent to quenched and tempered steels [1,9,11,13,
strength steels to reduce weight and fuel consumption of modern vehi­ 14]. The aging kinetics of these alloys to attain appropriate strength
cles. High specific strength steels are a novel avenue of research with levels is sluggish, typically taking between 30 and 120 h, and efforts
applications in automotive and structural lightweighting. Steels with have been made to accelerate the age hardening process to aid industrial
elevated concentrations of aluminum (>6 wt %), manganese (>20 wt scale-up [1,15,16]. These steels are an attractive candidate for weight
%), and carbon (0.6-2.0 wt %) termed Fe–Mn–Al–C steel, have been reduction in transportation due to the associated decrease in density and
developed to fit this need. Alloying with aluminum is performed for two an increase in the specific strength as compared to currently imple­
purposes: reduce the density and enable precipitation hardening by the mented materials.
decomposition of the supersaturated austenite (γ) to form κ-carbide Advanced characterization techniques have been applied to investi­
(FeMn)3AlC [1–3]. Elevated Mn, Al, and C levels are used to further gate the nano-scale κ-carbides. Transmission electron microscopy (TEM)
control the γ-austenite and κ-carbide stability. The density reduction has been used extensively to understand the shape, size, and orientation
observed in Fe–Mn–Al–C alloys, generally attributed to the Al-content relationship of the κ-carbide [1,17–19]. It has been shown that κ-carbide
on a linear basis of 1.4% density reduction per 1 wt % Al addition, is formation in high alloyed Fe–Mn–Al–C steels is a result of concomitant
attained by a combination of direct substitution for the heavier iron spinodal decomposition and ordering [1,20]. It has also been shown
atoms as well as a dilation of the lattice. Alloying with small amounts of from numerous TEM studies using selected area diffraction that the
Mo (0 to 0.7 wt%) has been shown to slow the age hardening rate while κ-carbide forms along a cube-on-cube orientation relationship or (100)γ
also increasing the impact toughness [4]. Interstitial C is also known to || (100)κ and [100]γ || [100]κ [1,17,20]. Atom probe tomography (APT)
dilate the matrix, and this was recently quantified in a Fe–30Mn-xC steel has been indispensable to understand the compositional distribution
to be on the order of +0.0029 Å per wt. % C [5]. These alloys typically within both γ-austenite and κ-carbide of the age hardened alloys [1,17,
exhibit high tensile strengths in excess of 1 GPa with total elongations 20,21]. Bartlett et al. [1] and Kim et al. [19] investigated the effect of Si
greater than 50% and Charpy v-notch impact toughness at room tem­ on partitioning of C and the subsequent impact on the strengthening of
perature exceeding 200 J in the solution treated condition [6,7]. The the Fe–Mn–Al–C steels. They both showed using APT that Si increased
precipitation age hardening treatment is typically performed between the partitioning of C into κ-carbide and that Si appeared to increase the
723 and 973 K to produce a homogeneously distributed coherent strength of the alloy at an equivalent aging time compared to lower or no

* Corresponding author.
E-mail address: marcus.young@unt.edu (M.L. Young).

https://doi.org/10.1016/j.msea.2021.141518
Received 10 September 2020; Received in revised form 6 May 2021; Accepted 7 May 2021
Available online 29 May 2021
0921-5093/© 2021 Elsevier B.V. All rights reserved.
N.A. Ley et al. Materials Science & Engineering A 820 (2021) 141518

Si. Along with the accelerated aging kinetics, Si was shown to reduce the solution-treated and water-quenched condition without any age-
austenite lattice parameter [1]. hardening are designated as “STQ”. Three different precipitation se­
The goal of this study is to identify the mechanisms for simulta­ quences were additionally performed on some of the STQ samples and
neously increasing strength and toughness in an age-hardened have been designated by their respective age-hardening process. One of
Fe–Mn–Al–C steel. To achieve this goal, the evolution of the the aging treatments followed the conventional single-stage isothermal
γ-austenite matrix and κ-carbide precipitates during various heat treat­ hold, while the other two applied the use of two-stage aging treatments.
ments and their effect on the impact toughness of a nominal The heat treatments and their designations are as follows: 1) “30”, a
Fe–30Mn–9Al–1C–1Si-0.7Mo alloy were investigated using scanning traditional 30 h hold at 803 K, 2) “2-2”, a 2 h hold at 723 K followed by 2
electron microscopy (SEM), TEM, synchrotron radiation x-ray diffrac­ h hold at 873 K, and 3) “4-1”, a 4 h hold at 723 K followed by 1 h at 923
tion (SR-XRD), hardness, and Charpy v-notch impact testing. An initial K. Samples were air-cooled after the precipitation heat treatment. So­
solution-treatment and water-quench and three subsequent age- lution treatment was performed to produce a single-phase supersatu­
hardening conditions are investigated in this study. In situ SR-XRD ex­ rated austenite. Subsequent heat treatment rational were as follows: the
periments were performed to track the dissolution and growth of the 30 condition was selected in reference to previously reported heat
κ-carbide precipitates. The relationships between microstructure, phases treatment to produce κ-carbide by precipitation hardening [2–17]. The
present, and impact toughness are explored. The experimental results 2-2 and 4-1 treatments were selected to reduce the aging time by
are compared with CALPHAD predictions. The results from this study nucleating particles at the first heat treatment stage, followed by growth
illuminate the combined strengthening and toughening mechanisms of the particles during the second stage of aging.
from age-hardening, due to the phase stability of the γ-austenite matrix
and κ-carbide precipitates in Fe–Mn–Al–C steels. 2.2. Material characterization

2. Experimental methods Heat-treated samples were characterized using SEM, TEM, hardness
testing, and Charpy impact testing at room temperature. The hardness
2.1. Material processing and impact property metrics were preferentially considered because
those two metrics, other than ballistic performance, are required for
A 190 kg ingot with a nominal composition of Fe–30Mn–9Al–1C–1Si- armor plate materials in accordance with MIL-DTL-12560K [26]. Plate
0.7Mo by weight was produced with a vacuum induction melting cross-sectional samples were prepared for SEM by mechanical polishing
furnace and cast into a 250 × 610 × 150 mm3 ingot. After casting, the down to 0.02 μm colloidal silica solution. SEM images were taken with a
ingot was coated with a ceramic wash to prevent oxidation and was Phenom XL operating at 15 kV using the backscatter mode. Hardness
reheated to 1423 K, and rolled to a plate thickness of 16 mm to produce a testing was conducted, after grinding of the sample surface to remove
total hot reduction of 89%. After the final rolling pass, the plate was any oxide or decarburization layer on the plate surfaces, using Rockwell
water-quenched to room temperature. The impurity content was hardness following ASTM E18 [27], with reported values having been
determined to be 0.01 wt % S and <0.005 wt % N and P, which is suf­ converted to Brinell in accordance with ASTM E140 [28]. Charpy impact
ficiently low to not expect a detrimental effect on the properties specimens were prepared from the mid-thickness region of the plate in
resulting from these types of impurities [22–25]. the transverse-longitudinal (T-L) with respect to the rolling direction,
Coupons of size 42 × 60 × 16 mm3 were obtained from the plate to following ASTM E23 [29]. Impact testing was performed on a Tinius
determine the toughness and microstructural response during heat Olsen Charpy impact machine outfitted with an instrumented striker to
treatment. Samples were sealed in stainless steel heat treatment bags to obtain the maximum load at fracture and total absorbed energy of the
mitigate oxidation and decarburization and were heat treated as shown sample.
schematically in Fig. 1. All samples underwent a 2 h solution treatment The TEM specimens were prepared from the aged samples by
at 1323 K followed by water-quenching. Samples that remained in this punching out 3 mm discs, which were subsequently ground to roughly
100 μm in thickness. A Gatan Model 656 dimple grinder was then used
to generate a dimple with a final thickness of ≤10 μm within the thinned
discs. After dimpling, ion milling of the discs was performed with a
Gatan 695 Precision Ion Polishing System (PIPS) until an electron
transparent perforation formed in the dimpled region. The perforated
specimens were then analyzed in a JEOL 2100F (scanning) TEM ([S]
TEM) operated at 200 kV in both STEM and diffraction mode.
The phase structure for each of the samples was determined using
high energy SR-XRD at the Advanced Photon Source on the 11-ID-C
beamline at Argonne National Laboratory. The general setup for the
SR-XRD experiments is similar to those found in Refs [30,31]. The
samples examined in the ex situ SR-XRD experiments were in the STQ,
30, 2-2, and 4-1 heat treatment conditions and involved multiple mea­
surements at different locations on the sample at room temperature,
where only one representative measurement is presented here and was
typically taken from the center of the sample. The samples examined in
the in situ SR-XRD experiments were separated into two categories:
dissolution and growth. The dissolution samples started in the 2-2, 4-1,
and 30 heat treatment conditions at room temperature. The growth
sample started in the STQ heat treatment condition at room tempera­
ture. DeBye-Scherrer diffraction patterns were collected using a mono­
Fig. 1. Schematic of the Fe–Mn–Al–C heat treatments. Each sample underwent
an initial 2 h solution treatment at 1323 K followed by a water-quench, while chromated 105.5 keV (λ = 0.117300 Å) X-ray beam with a rectangular
some of the samples were further age-hardened followed by air-cooling as beam with a size of 200 × 200 μm2. Diffraction images were collected
follows: 1) “30”, a traditional 30 h hold at 803 K, 2) “2-2”, a 2 h hold at 723 K using a PerkinElmer amorphous silicon detector which was positioned
followed by 2 h hold at 873 K, and 3) “4-1”, a 4 h hold at 723 K followed by 1 h behind the beam stop and approximately 1.50 m from the sample. The
at 923 K. integration of the full 2-D DeBye-Scherrer diffraction patterns to 1-D

2
N.A. Ley et al. Materials Science & Engineering A 820 (2021) 141518

lineouts involved converting the patterns from polar coordinates to γ-austenite is measured to be 245 ± 70 μm and is unaffected by the low
Cartesian coordinates, then integrating the full pattern to a single line­ temperature aging treatments. consistent with the work by Field et al.
out, and finally normalizing the lineout by dividing the intensity mea­ [7] that showed negligible grain growth after 15 h at 1173 K. Annealing
surements with the most intense diffraction peak, which corresponded twins are also readily observed in the micrographs due to the high so­
to the γ-austenite (111) peak in all cases, and plotting the normalized lution treatment temperature and the large grain size, which increases
intensity as a function of d-spacing. Calibration of the SR-XRD experi­ the probability of forming twins.
ments was performed using a standard Ce2O powder. Each sample was TEM analysis was performed to further investigate the grain
positioned, so that the center of the sample was in the center of the boundaries for the presence of kappa precipitates or impurities and grain
beam. The beam exposure time was approximately 0.1 s and measure­ interiors for verification of κ-carbide formation. STEM dark field images
ments were averaged over 10 frames, for a total exposure time of 1 s per and selected area electron diffraction (SAED) patterns of the 2-2 and 4-1
sample. In situ heating and cooling of the samples was performed using a samples are shown in Figs. 4 and 5, respectively. In dark field images for
Linkam T1500 temperature-controlled stage that had a constant flow of both conditions, a grain boundary between two austenite grains is pre­
Ar gas into the furnace to prevent oxidation. The dissolution samples sent as seen by the stark diffraction contrast difference between the two
were all heated from room temperature (~300 K) to 1323 K at a rate of respective grains. Additionally, the distinct contrast modulation within
10 K/min and then held for 1 h. The growth sample was heated from each grain is due to the high volume fraction of κ-carbide present. The
room temperature (~300 K) to 723 K and held for 4 h, then heated to bright primary reflections within each SAED were indexed as
923 K and held for 1 h, with a heating rate of 10 K/min. A schematic for γ-austenite, while the dim superlattice reflections correspond to the
each of the in situ heat treatments can be seen in Fig. 2. κ-carbide and its perovskite crystal structure. The γ-austenite lattice
parameters for the 2-2, 4-1, and 30 conditions were 3.716 ± 0.015 Å,
2.3. Thermodynamic and kinetic Modeling 3.735 ± 0.034 Å, and 3.767 ± 0.022 Å, while the kappa carbide lattice
parameters were 3.693 ± 0.053 Å, 3.773 ± 0.045 Å, and 3.773 ± 0.035
Thermodynamic and kinetic predictions following the calculation of Å, respectfully. This equates to a lattice misfit of − 0.61% (compressive),
phase diagrams (CALPHAD) method were performed to evaluate the 1.01% (tensile), and 0.16% (tensile) for the 2-2, 4-1, and 30 conditions,
precipitation and dissolution of the κ-carbide precipitates. The ther­ respectfully. These values are higher than other reported values
modynamic equilibrium phase constitution for the nominal alloy although the misfit is comparable except for the apparent tensile misfit
composition was predicted using ThermoCalc 2020a with the TCFE9 which has not been previously reported experimentally [1,19]. Because
database. Precipitation kinetics of κ-carbide in an austenite matrix were of the large overlapping standard deviations the significance of the
also simulated using the PRISMA module with the MOBFE2 mobility values is minimized, and they are only here reported and not the basis of
database appended. All calculations were performed for 1 mol of ma­ the conclusions drawn.
terial at 1 atm pressure and the cubic β-Mn phase was removed since it Samples in the STQ and aged conditions were analyzed using SR-XRD
has not been observed to form in these Si containing Fe–Mn–Al–C alloys. to provide correlation of the phase structure with the microscopy and
The precipitation of κ-carbide was modeled as a bulk precipitation re­ impact toughness. Fig. 6a–d shows ex situ integrated SR-XRD patterns
action in a cuboidal morphology in agreement with TEM observations plotting normalized intensity vs. d-spacing as well as the full DeBye-
and phase field simulations of κ-carbide and its growth in comparable Scherrer diffraction pattern for the Fe–Mn–Al–C steel in the STQ, 2-2,
Fe–Mn–Al–C alloys aged to similar conditions [32–35]. 4-1, and 30 heat treatment conditions, respectively.
The DeBye-Scherrer diffraction patterns from all four conditions
3. Results & discussion show large diffractions spots due to the large γ-austenite grains present
in the samples. A minimal amount of texture, i.e. preferentially oriented
3.1. Microscopy grains, is observed in the full DeBye-Scherrer diffraction patterns for all
four conditions, in this case, due to the rolling process, as evident by the
SEM backscattered electron imaging of the aged samples, shown in symmetry of diffraction spots in any given γ-austenite diffraction plane.
Fig. 3, demonstrate that the alloy appears to be primarily composed of The predominant phase in all of the diffraction patterns shown in Fig. 7
γ-austenite. The κ-carbide is not observed to have formed at grain is the γ-austenite matrix, which exhibits a face-centered cubic crystal
boundaries and there is no visible δ-ferrite. The grain size of the structure and is consistent with the microscopy results. All of the ex situ
samples also contain κ-carbide, including the STQ sample indicating the
quench rate which was insufficient to prevent κ-carbide formation
during cooling. Accurate phase quantification for these samples was not
possible because the large grain size resulted in spotty diffraction pat­
terns rather than precise rings. The combination of spotty patterns and
small diffraction peaks makes an accurate quantification of the κ-carbide
in each sample imprecise. Furthermore, a direct quantitative compari­
son between samples is not possible as even small variations in the
sample sizes affect the total volume sampled. In lieu of quantitative
analysis, the authors have qualitatively ranked the κ-carbide volume
fraction based on normalized intensity as 30 > 4-1 ≥ 2-2 > STQ. The
aged samples (2-2, 4-1, and 30) further have sidebands around the
γ-austenite (200) peak characteristic in diffraction patterns of face-
centered cubic alloys undergoing spinodal decomposition. These side­
bands result from the periodic concentration fluctuations giving rise to
modulation of the lattice parameter spacing conditions and are
comprised of κ-carbide (larger d-spacing) and the solute depleted
γ-austenite, denoted by γ* (smaller d-spacing). The intensity of the
sidebands was different based on the heat treatment the samples had
Fig. 2. Temperature-time profiles for the SR-XRD experiments during in situ experienced. Increased sideband intensity at the expense of the parent
heating of a Fe–Mn–Al–C steel for the dissolution and growth heat treatments, peak is indicative of enhanced ordering and chemical partitioning giving
respectively. rise to two distinct regions. Sidebands were not observed in the STQ

3
N.A. Ley et al. Materials Science & Engineering A 820 (2021) 141518

Fig. 3. SEM-BSE images of the heat treated alloy in the (a) 2-2, (b) 4-1, and (c) 30 conditions, respectively.

Fig. 4. STEM dark field image of a grain boundary in the “2-2” sample and insets of the SAED patterns taken along the [111] and [112] zone axes.

Fig. 5. STEM dark field image of a grain boundary in the “4-1” sample and insets of the SAED patterns taken along the [111] and [112] zone axes.

sample as is consistent with weak ordering without chemical partition­ that of the parent supersaturated γ-austenite (200) peak. The sidebands
ing. The most pronounced sidebands were visible for the single-stage observed for sample 2-2 were also distinct peaks although their relative
heat treatment 30, in which the sideband intensity was greater than intensities were less than that of the parent peak. Sample 4-1 had

4
N.A. Ley et al. Materials Science & Engineering A 820 (2021) 141518

Fig. 6. Integrated ex situ SR-XRD diffraction patterns plotting normalized intensity vs d-spacing for a Fe–Mn–Al–C steel in the: a) STQ, b) 2-2, c) 4-1, and d) 30 heat
treat conditions. The insets show the full DeBye-Scherrer 2-D diffraction patterns for each condition, respectively.

peak begins to slowly decrease as the samples are heated up from room
temperature to some critical temperature before a rapid dissolution of
the κ-carbide phase occurs. During this rapid dissolution, multiple in­
flection points in the line can be observed and are likely due to a change
in morphology as the κ-carbide phase dissolves. Eventually the κ-carbide
phase is fully dissolved. The critical temperature before rapid dissolu­
tion were measured as 773 K (30), 873 K (2-2), and 903 K (4-1). The
temperature at which full dissolution is reached were measured as 1023
K (30), 1088 K (2-2), and 1123 K (4-1). The time it takes from the onset
of rapid dissolution to full dissolution is comparable (~1000 s) for all
three conditions; however, the highest dissolution onset and termination
temperatures were measured in the 4-1 + STQ condition compared to
the 30 + STQ condition, which had the lowest onset and termination
temperatures. This result suggests the κ-carbide phase the 4-1 + STQ
condition is more stable than that in the 30 + STQ condition and the 2-2
+ STQ condition. Although not shown here, similar results for the
κ-carbide (110) peak were also observed, with critical temperatures
occurring at the same time as the κ-carbide phase (100) peak.
In situ κ-carbide nucleation and growth were measured by tracking
Fig. 7. Time vs. normalized intensity of the (100) κ-carbide peak from in situ
the (100) and (110) κ-carbide peaks during heat treatment of an STQ
dissolution SR-XRD experiments for a Fe–Mn–Al–C steel in the: a) 2-2 + STQ
(red), b) 4-1 + STQ (green), and c) 30 + STQ (black) conditions, respectively. sample according to condition 4-1. It can be seen in Fig. 8 that a small
(For interpretation of the references to colour in this figure legend, the reader is amount of κ-carbide is already present at the start of the in situ growth
referred to the Web version of this article.) SR-XRD experiment (at time equals zero and room temperature) as was
observed in the ex situ analysis. A slight inflection exists in the early
shoulders on the γ-austenite (200) peak indicative of the sidebands, seconds of the experiment suggesting that some nucleation is still
although the definition was significantly weaker suggesting a lesser occurring prior to growth which begins around 40 min or just over 673
degree of chemical partitioning. K. A constant growth of κ-carbide phase occurs during the 4 h hold at
The dissolution of κ-carbide was measured for each of the aged 723 K, as evident by the basically linear slope of the line in Fig. 7 for both
conditions by tracking the (100) κ-carbide peak, which was chosen due the (100) and (110) κ-carbide peaks and then accelerates as the tem­
to the isolation of the peak so that no neighboring peaks would interfere perature is increased and eventually held for 1 h at 923 K.
with the data analysis. Fig. 7 shows the in situ dissolution of the (100)
κ-carbide peak as a function of time for the 2-2 + STQ (red), 4-1 + STQ
(green), and 30 + STQ (black) conditions, respectively, with key tem­ 3.2. CALPHAD predictions
peratures indicated. In all three conditions, the intensity of the κ-carbide
The equilibrium thermodynamic phases predicted using CALPHAD

5
N.A. Ley et al. Materials Science & Engineering A 820 (2021) 141518

Fig. 9. Simulated time-temperature-transformation (TTT) diagram for bulk


precipitation of κ-carbide (black) and M6C (blue). The series of curves shown
represents the predicted time at temperature when each precipitate reaches
volume fractions of 0.001 (dotted), 0.01 (dashed), and 0.1 (solid). (For inter­
pretation of the references to colour in this figure legend, the reader is referred
to the Web version of this article.)
Fig. 8. Time vs. normalized intensity of the (100) κ-carbide peak in situ for a
Fe–Mn–Al–C Steel in the STQ + 4-1 heat treatment condition.

Table 1
are generally consistent with observation. A high temperature single- Simulated precipitation during quenching of Fe–Mn–Al–C steel from 1323 K to
phase austenite region exists from 1205 to 1312 K. Upon cooling from 723 K over a range of cooling rates.
the austenitic phase field, two carbides are predicted to exist: M6C
Cooling Rate Temperature for 0.001 Vf Vf at 723 K
((Fe2Mo)2(Fe,Mo,Si)2C1) beginning at 1205 K and increasing to an
[K/s] M6C M6C
equilibrium mole fraction at 723 K of 1.2%, and κ-carbide ((Fe, κ-carbide κ-carbide

Mn)3Al1(C,Va)1) starting at 1136 K and rapidly increasing to an equi­ 0.1 1047.6 901.4 0.1073 0.0028
librium mole fraction of 19.1% at 723 K M6C has only been experi­ 1 1033.0 – 0.1029 7 x 10− 6
10 1007.9 – 0.0647 7 x 10− 10
mentally observed in Mo bearing Fe–Mn–Al–C steels with Mo content 100 – – 0.0002 5 x 10− 27
greater than 3 wt % [36,37] due in large part to sluggish Mo diffusion, 1000 – – 2 x 10− 15 –
although the effect of Si content on M6C formation has not been eval­
uated. Finally, α-ferrite is predicted by CALPHAD to also become stable
below 1045 K, reaching an equilibrium mole fraction of 23.8% at 723 K. starting from a fully solution treated condition and returning to 298 K
Because of the diffusive transformation required to form α-ferrite at low before reheating to 1323 K for dissolution. A heating and cooling rate of
temperatures, the kinetics prevent this equilibrium phase from forming. 10 K/min was used for all heating and cooling segments, in accordance
In the alloy under investigation, α-ferrite is not experimentally observed with the in-situ SR-XRD experiments, and the κ-carbide was considered
even after 30 h at 803 K, which is indicative of the formation rate. as bulk precipitation with a cuboidal morphology. The precipitation
A time-temperature-transformation (TTT) diagram was simulated simulation results (Table 2) show that the temperature at which the
considering bulk precipitation of cuboidal κ-carbide and spherical M6C volume fraction exceeds 0.001 is the lowest for aging treatment 4-1 and
within the γ-austenite matrix. A series of curves was generated for vol­ highest for 30. Correspondingly, the size of the precipitates at 0.001 Vf
ume fractions of 0.001, 0.01, and 0.1 up to a 120,000 s isothermal hold, was smaller when this volume fraction was attained at a lower tem­
as shown in Fig. 9. It can be seen that the M6C kinetics are significantly perature. This is in good agreement with the convention of the two-stage
slower than κ-carbide and although the equilibrium stability of M6C was aging concept in which a more homogenous distribution of fine pre­
predicted to be higher than κ-carbide, the nose of the TTT curve is 60 K cipitates is nucleated at the low temperature hold prior to stepping up to
greater for κ-carbide. a higher temperature for accelerated growth. By holding the sample for
Experimental solution treating and water quenching of the an additional 2 h at 723 K before ramping to the second aging stage, the
Fe–Mn–Al–C alloy was intended to produce a fully austenitic micro­ average initial precipitate size of the 4-1 treatment was smaller than 2-2.
structure with all alloying elements in solid solution, however the SR- The final volume fraction of κ-carbide is lowest for treatment 4-1 and
XRD results indicated a small amount of κ-carbide present in the STQ highest for 30, with the average length of the κ-carbide largest for 4-1
sample. Kinetic simulations were conducted for κ-carbide and M6C and smallest for 30. Both two stage heat treatment simulations pro­
precipitation during quenching of the nominal alloy from the solution duced larger precipitates but a smaller total volume compared to the
treatment temperature of 1323 K–723 K over a range of cooling rates traditional single stage precipitation treatment. The final average
from 0.1 K/s to 1000 K/s. The simulation results shown in Table 1 composition of the κ-carbide for each simulated heat treatment is also
confirm that κ-carbide precipitation requires increasing amounts of shown in Table 2. The Mn-content of the κ-carbide is 5.0% and 6.9%
undercooling to be physically realized compared to the initial equilib­ lower in the two-stage aged 2-2 and 4-1 conditions, respectively, than in
rium stability at 1136 K for faster cooling rates. The cooling rate of the the conventionally aged 30 condition. The degree of carbon off-
STQ samples to form a small but measurable amount of κ-carbide is stoichiometry is increased for both of the two-staged aged samples,
predicted to be between 10 and 100 K/s in agreement with conventional although significantly smaller than the experimentally measured carbon
cooling rates associated with water-quenching. concentrations that are generally on the order of 10 at.%, or about 50%
The precipitate growth sequence and subsequent dissolution of carbon vacancies [38]. The Al-content in all three conditions is similar.
κ-carbide were simulated for each of the heat treatment conditions Because Si and Mo are forbidden in the κ-carbide thermodynamic model,

6
N.A. Ley et al. Materials Science & Engineering A 820 (2021) 141518

Table 2
Characteristics of simulated κ-carbide precipitation in an Fe–Mn–Al–C alloy for three heat treatment profiles: the temperature and average precipitate size when the
treatment first formed 0.001 Vf κ-carbide, the final volume fraction, average length, and composition of the precipitated κ-carbide.
Heat Treatment Profile 0.001 Vf Temperature [K] 0.001 Vf Average Size [nm] Final Vf Final Average Length [nm] Fe (at %) Mn (at %) Al (at %) C (at %)

30 784.2 0.735 0.1679 4.368 16.045 43.955 20.000 19.999


2–2 768.9 0.707 0.1580 5.987 18.237 41.767 20.001 19.995
4–1 729.3 0.657 0.1558 7.698 19.074 40.932 20.002 19.991

the experimentally observed partitioning of Si is not able to be captured


within these models. Further, the simulated chemical segregation is
insufficient to predict meaningful changes in the matrix lattice
parameter.
The subsequent dissolution simulations completely dissolved the
κ-carbide around 1140 K regardless of the initial aging treatment;
however, the critical dissolution temperature at which the rate of
dissolution accelerated was correlated to the simulated precipitate size.
Aging treatment 30 (finest κ-carbide) had the lowest critical dissolution
temperature and 4-1 (coarsest κ-carbide) had the highest. This result
shown in Fig. 10 is in good agreement with the SR-XRD observations
(Fig. 6) regarding the trends, although the temperature range was larger
for the experimental measurement than in the CALPHAD predictions.
The SR-XRD measured critical dissolution temperatures from the CAL­
PHAD simulated values by: TSR-XRD
crit = 3.8921 * TCALPHAD
crit – 2401.3.

3.3. Impact toughness

The impact toughness of the samples was evaluated using instru­


mented impact testing on standard Charpy V-notch bars at room tem­ Fig. 11. Charpy V-Notch absorbed energy as a function of hardness for
perature (22 ± 1 ◦ C). The total energy absorbed is shown in Fig. 11 as a Fe–Mn–Al alloy following various heat treatments.
function of the measured hardness. The STQ condition was the softest
(250 ± 13) and had a significantly higher impact toughness of 174.0 ±
9.0 J, which was not shown for better clarity of the aged condition re­
sponses. There is an increase in absorbed energy in from the 30 to the 2-
2 at equivalent hardness. Additionally, the 4-1 condition increases in
both hardness and toughness compared to the conventional 30 condi­
tion, which is an atypical material response.
The SR-XRD results and CALPHAD predictions showed that the two-
stage aging treatments produced a smaller volume fraction of coarser
κ-carbide than the conventionally aged sample, although they achieved
equivalent or higher hardness. κ-carbide has previously been shown to
be a weak dislocation barrier [39]; therefore, by attaining the required
hardness with a smaller volume fraction of κ-carbide, the degree of glide
plane softening was reduced and the solid solution strengthening was
retained. The correlation between Charpy impact toughness and the
κ-carbide volume fraction of the three aging conditions is shown in
Fig. 12, using CALPHAD predicted volume fractions for quantitative

Fig. 12. Charpy V-Notch impact energy as a function of CALPHAD predicted


κ-carbide volume fraction for Fe–Mn–Al alloy following various
heat treatments.

analysis. It should be noted that the STQ condition followed the linear
trend observed in Fig. 12 with essentially no κ-carbide present and the
highest impact toughness, although it was not shown for better clarity of
the aged condition responses. An additional unquantified correlation
with toughness is the degree of ordering within the precipitated κ-car­
bide represented by the SR-XRD sidebands. Impact toughness decreased
with increasing sideband definition and intensity, which may be due in
Fig. 10. Volume fraction of κ-carbide as a function of temperature during part to increased volume fraction, but more significantly on chemical
simulated dissolution at 10 K/s indicating the temperature for each starting modulation. Additional analysis such as atom probe tomography would
condition where the dissolution accelerates.

7
N.A. Ley et al. Materials Science & Engineering A 820 (2021) 141518

be required to quantify this effect, and is a topic of further investigation. condition exhibited the highest hardness and absorbed the highest
The hardness, a proxy for strength, does not correlate to the amount amount of energy, i.e. the most toughness. This is a key result as most
of κ-carbide formed and the highest hardness occurs with the lowest materials do not exhibit increased hardness with an increase of
κ-carbide intensity in the 4-1 condition, as well as with the lowest toughness and is likely due to the combination of κ-carbide pre­
sideband formation. This would imply that the high degree of disorder cipitates and almost no depleted austenite (γ*). Thus, the key finding
and reduced sideband formation provides a greater strengthening to the is that the high hardness and high toughness observed in the 4-1
alloy compared to the precipitation of the κ-carbide alone as seen in the condition is likely due to the reduced amount of spinodal decom­
30 and 2-2 conditions. This type of behavior is in good agreement with position of γ-austenite phase, as well as the strengthening presence
previous researchers noting that the κ-carbide is a weak, shearable and stability of the κ-carbide nano-precipitates.
dislocation barrier [30]; evidence suggests that upon reaching a critical
radius (~10–13 nm) κ-carbide may begin to act as an Orowan barrier Author Credit Statement
[19,35,40]. Because the average predicted length of κ-carbide is under
this value for all heat treatments considered, Orowan strengthening is All persons who meet authorship criteria are listed as authors, and all
not expected to have contributed to the observed increase in hardness authors certify that they have participated sufficiently in the work to
for the 4-1 condition, or any conditions examined in this work. The take public responsibility for the content, including participation in the
traditionally aged 30 and the 2-2 conditions would illustrate the loss of concept, design, analysis, writing, or revision of the manuscript.
hardness or strength associated with the γ* ordering reaction which Furthermore, each author certifies that this material or similar material
would support the concept that solid solution strengthening of the has not been and will not be submitted to or published in any other
austenite in conjunction with some κ-carbide provides the best combi­ publication before its appearance in Materials Science and Engineering A.
nation of strength and toughness. Acknowledgements All persons who have made substantial contri­
butions to the work reported in the manuscript (e.g., technical help,
4. Conclusions writing and editing assistance, general support), but who do not meet
the criteria for authorship, are named in the Acknowledgements and
In this study, the toughness and phase stability of the γ-austenite have given us their written permission to be named. If we have not
matrix, solute depleted austenite (γ*) phase, and κ-carbide precipitates included an Acknowledgements, then that indicates that we have not
in an Fe–30Mn–9Al–1C–1Si-0.7Mo alloy were investigated. Four con­ received substantial contributions from non-authors.
ditions were evaluated: an initial solution treatment at 1323 K for 2 h
followed by water-quenching (STQ) and three additional age-hardening
treatments, which consisted of a traditional 30 h hold at 803 K (30), a 2 h Declaration of competing interest
hold at 723 K followed by 2 h hold at 873 K (2-2), and a 4 h hold at 723 K
followed by 1 h at 923 K (4-1). In addition to conventional character­ The authors declare that they have no known competing financial
ization using SEM, TEM, hardness testing, and Charpy impact testing, ex interests or personal relationships that could have appeared to influence
situ and in situ SR-XRD experiments were performed, where the latter the work reported in this paper.
included three dissolution and one growth experiments designed to
specifically monitor the κ-carbide precipitate response. The experi­ Acknowledgements
mental results were compared with CALPHAD simulations. Based on this
study, the following conclusions can be made: The authors would like to thank Yang Ren, beam line scientist, for
helping with the experiments at the Advanced Photon Source (APS).
1) The predominant phase for each condition was relatively large- This research used resources of the Advanced Photon Source, a U.S.
grained γ-austenite with annealing twins. The STQ condition also Department of Energy (DOE) Office of Science User Facility operated for
showed a very small amount of κ-carbide precipitates consistent with the DOE Office of Science by Argonne National Laboratory under Con­
CALPHAD predicted quench rates between 10 and 100 K/s. The 30 tract No. DE-AC02-06CH11357.
condition showed a small amount of κ-carbide precipitates and a
small amount of depleted γ*-austenite which resulted from spinodal
References
decomposition of the γ-austenite. The 2-2 condition showed about
the same small amount of κ-carbide precipitates and an even smaller [1] L.N. Bartlett, D.C. Van Aken, J. Medvedeva, D. Isheim, N.I. Medvedeva, K. Song, An
amount of depleted γ*-austenite, indicating less spinodal decompo­ atom probe study of kappa carbide precipitation and the effect of silicon addition,
sition of the γ-austenite, as compared to the 30 condition. Finally, the Metall. Mater. Trans. 45 (5) (2014) 2421–2435.
[2] G. Frommeyer, U. Brüx, Microstructures and mechanical properties of high-
4-1 condition showed about the same small amount of κ-carbide strength Fe-Mn-Al-C light-weight TRIPLEX steels, Steel Res. Int. 77 (9-10) (2006)
precipitates and almost no amount of depleted γ*-austenite as 627–633.
compared to the 30 and 2-2 conditions. The CALPHAD simulated [3] H. Kim, D.-W. Suh, N.J. Kim, Fe-Al-Mn-C lightweight structural alloys: a review on
the microstructures and mechanical properties, Sci. Technol. Adv. Mater. 14 (1)
volume fractions of κ-carbide agreed with this experimental (2013).
observation. [4] T. Constance, L. Bartlett, R. Vaz Penna, AFS/FEF student technology contest:
2) Based on the in situ dissolution SR-XRD experiments, the 4-1 + STQ understanding the role of Mo on к-carbide precipitation and fracture of lightweight
Advanced high-strength steels, Int. J. Metalcast. 13 (2018).
condition shows the most stable κ-carbide precipitates, which [5] Y. Tomota, T. Murakami, Y.X. Wang, T. Ohmura, S. Harjo, Y.H. Su, T. Shinohara,
require higher temperatures to initiate and fully complete dissolu­ Influence of carbon concentration and magnetic transition on the austenite lattice
tion, while the 30 + STQ condition shows the earliest initiation and parameter of 30Mn-C steel, Mater. Char. 163 (2020), 110243.
[6] D.M. Field, K.R. Limmer, Effect of Solution Treatment on Grain Size and Toughness
completion of dissolution of the κ-carbide precipitates. The CAL­ of Lightweight Fe-Mn-Al-C Steel, AISTech 2019 — Proceedings of the Iron & Steel
PHAD simulations further predicted that this reduced stability is the Technology Conference, Association for Iron & Steel Technology, Pittsburgh, PA,
result of forming a larger volume of smaller precipitates with USA, 2019.
[7] D.M. Field, K.R. Limmer, B.C. Hornbuckle, On the grain growth kinetics of a low
increased Mn-partitioning to the κ-carbide compared to both of the
density steel, Metals 9 (9) (2019) 997.
two-stage treatments. [8] O. Acselrad, R.A. Simao, L.C. Pereira, C.A. Achete, I.S. Kalashnikov, E.M. Silva,
3) The κ-carbide stability was shown to have implications for the Phase transformations in FeMnAlC austenitic steels with Si addition, Metall. Mater.
impact toughness; both the 30 and 2-2 conditions showed the same Trans. 33 (11) (2002) 3569–3573.
[9] R. Howell, T. Weerasooriya, D.C. Van Aken, Tensile, High Strain Rate Compression
hardness, however the 2-2 condition demonstrated a higher tough­ and Microstructural Evaluation of Lightweight Age Hardenable Cast Fe-30Mn-9Al-
ness in the Charpy V-notch absorbed energy. Moreover, the 4-1 XSi-0.9C-0.5Mo Steel, AFS Transactions, 2009, pp. 751–763.

8
N.A. Ley et al. Materials Science & Engineering A 820 (2021) 141518

[10] I.S. Kalashnikov, O. Acselrad, A. Shalkevich, L.D. Chumakova, L.C. Pereira, Heat [26] Department of the Army, MIL-DTL-12560K, Detail Specification: Armor Plate,
treatment and thermal stability of FeMnAlC alloys, J. Mater. Process. Technol. 136 Steel, Wrought, Homogeneous (For Use in Combat-Vehicles and for Ammunition
(1) (2003) 72–79. Testing, 2013.
[11] G. Kayak, Fe− Mn− Al precipitation-hardening austenitic alloys, Met. Sci. Heat [27] ASTM International, Standard Test Methods for Rockwell Hardness of Metallic
Treat. 11 (2) (1969) 95–97. Materials, ASTM E18-16, West Conshohocken, PA, 2016.
[12] B. Hallstedt, A.V. Khvan, B.B. Lindahl, M. Selleby, S. Liu, PrecHiMn-4—a [28] ASTM International, Standard Hardness Conversion Tables for Metals Relationship
thermodynamic database for high-Mn steels, Calphad 56 (2017) 49–57. Among Brinell Hardness, Vickers Hardness, Rockwell Hardness, Superficial
[13] Y.G. Kim, Y.S. Park, J.K. Han, Low temperature mechanical behavior of Hardness, Knoop Hardness, Scleroscope Hardness, and Leeb Hardness, ASTM E140-
microalloyed and controlled-rolled Fe-Mn-Al-C-X alloys, Metallurgical 12be1, West Conshohocken, PA, 2013.
Transactions A 16 (9) (1985) 1689–1693. [29] ASTM International, Standard Test Methods for Notched Bar Impact Testing of
[14] S.A. Buckholz, D.C. Van Aken, L.N. Bartlett, On the influence of aluminum and Metallic Materials, ASTM E23-16b, West Conshohocken, PA, 2016.
carbon on abrasion resistance of high manganese steels, AFS Transactions 120 (12) [30] M. Young, J. Almer, M. Daymond, D. Haeffner, D. Dunand, Load partitioning
(2013) 54. between ferrite and cementite during elasto-plastic deformation of an ultrahigh-
[15] K. Limmer, D. Field, B.A. Cheeseman, J.S. Montgomery, K.M. Sebeck, R.A. Howell, carbon steel, Acta Mater. 55 (6) (2007) 1999–2011.
Accelerated heat treatment of high-mn triplex steels for armor applications, Iron [31] M. Daymond, M. Young, J. Almer, D. Dunand, Strain and texture evolution during
Steel Technol. 15 (2018) 92–97. mechanical loading of a crack tip in martensitic shape-memory NiTi, Acta Mater.
[16] L. Bartlett, D. Van Aken, High manganese and aluminum steels for the military and 55 (11) (2007) 3929–3942.
transportation industry, JOM 66 (9) (2014) 1770–1784. [32] A. Rahnama, R. Dashwood, S. Sridhar, A phase-field method coupled with
[17] J. Moon, S.-J. Park, J.H. Jang, T.-H. Lee, C.-H. Lee, H.-U. Hong, H.N. Han, J. Lee, B. CALPHAD for the simulation of ordered κ-carbide precipitates in both disordered γ
H. Lee, C. Lee, Investigations of the microstructure evolution and tensile and α phases in low density steel, Comput. Mater. Sci. 126 (2017) 152–159.
deformation behavior of austenitic Fe-Mn-Al-C lightweight steels and the effect of [33] I. Gutierrez-Urrutia, D. Raabe, Influence of Al content and precipitation state on
Mo addition, Acta Mater. 147 (2018) 226–235. the mechanical behavior of austenitic high-Mn low-density steels, Scripta Mater.
[18] O.A. Zambrano, A general perspective of Fe-Mn-Al-C steels, J. Mater. Sci. 53 (20) 68 (6) (2013) 343–347.
(2018) 14003–14062. [34] A. Rahnama, H. Kotadia, S. Sridhar, Effect of Ni alloying on the microstructural
[19] C.W. Kim, M. Terner, J.H. Lee, H.U. Hong, J. Moon, S.J. Park, J.H. Jang, C.H. Lee, evolution and mechanical properties of two duplex light-weight steels during
B.H. Lee, Y.J. Lee, Partitioning of C into κ-carbides by Si addition and its effect on different annealing temperatures: experiment and phase-field simulation, Acta
the initial deformation mechanism of Fe-Mn-Al-C lightweight steels, J. Alloys Mater. 132 (2017) 627–643.
Compd. 775 (2019) 554–564. [35] J. Lee, S. Park, H. Kim, S.-J. Park, K. Lee, M.-Y. Kim, P.P. Madakashira, H.N. Han,
[20] I.S. Kalashnikov, O. Acselrad, L.C. Pereira, T. Kalichak, M.S. Khadyyev, Behavior of Simulation of κ-carbide precipitation kinetics in aged low-density Fe–Mn–Al–C
Fe-Mn-Al-C steels during cyclic tests, J. Mater. Eng. Perform. 9 (3) (2000) 334–337. steels and its effects on strengthening, Met. Mater. Int. 24 (4) (2018) 702–710.
[21] L.N. Bartlett, D.C. Van Aken, J. Medvedeva, D. Isheim, N. Medvedeva, K. Song, An [36] H. Oikawa, Review on Lattice Diffusion of Substitutional Impurities in Iron, A
atom probe study of κ-carbide precipitation in austenitic lightweight steel and the summary report, 1982.
effect of phosphorus, Metall. Mater. Trans. 48 (11) (2017) 5500–5515. [37] T.F. Liu, S.W. Peng, Y.L. Lin, C.C. Wu, Orientation relationships among M23C6,
[22] L. Bartlett, R. Rahman, A. Torres, Minimizing phosphorus pickup during melting M6C, and austenite in an Fe-Mn-Al-Mo-C alloy, Metallurgical Transactions A 21 (2)
and casting of lightweight Fe–Mn–Al–C steels, Int. J. Metalcast. 12 (2017) 1–18. (1990) 567–574.
[23] R. Howell, D. Aken, Microstructural and fracture behavior of phosphorus- [38] M.J. Yao, P. Dey, J.B. Seol, P. Choi, M. Herbig, R.K.W. Marceau, T. Hickel,
containing Fe-30Mn-9Al-1Si-0.9C-0.5Mo alloy steel, Metall. Mater. Trans. 46 J. Neugebauer, D. Raabe, Combined atom probe tomography and density
(2015). functional theory investigation of the Al off-stoichiometry of κ-carbides in an
[24] A. Svyazhin, V. Bazhenov, L. Kaputkina, I. Smarygina, V. Kindop, Nitrogen in austenitic Fe–Mn–Al–C low density steel, Acta Mater. 106 (2016) 229–238.
Fe–Mn–Al–C–based steels, CIS Iron and Steel Review 12 (2016) 13–17. [39] Z. Wang, W. Lu, H. Zhao, J. He, K. Wang, B. Zhou, D. Ponge, D. Raabe, Z. Li,
[25] D.T. Pierce, J.A. Jiménez, J. Bentley, D. Raabe, C. Oskay, J.E. Wittig, The influence Formation mechanism of κ-carbides and deformation behavior in Si-alloyed
of manganese content on the stacking fault and austenite/ε-martensite interfacial FeMnAlC lightweight steels, Acta Mater. 198 (2020) 258–270.
energies in Fe–Mn–(Al–Si) steels investigated by experiment and theory, Acta [40] I. Gutiérrez-Urrutia, D. Raabe, High strength and ductile low density austenitic
Mater. 68 (2014) 238–253. FeMnAlC steels: simplex and alloys strengthened by nanoscale ordered carbides,
Mater. Sci. Technol. 30 (9) (2014) 1099–1104.

You might also like