Assessment of the compositional requirements to form Fe-Mn-C austenite-martensite composites

1 Assessment of the compositional requirements to form Fe-Mn-C austenite-martensite composites Marcel Muencha, Reza Gholizadehb, Myeong-heom Parkb, Nobuhiro Tsujib, Martin Peterlechnerc, Yolita M. Eggelerc, Jan L. Riedela, Michael K. Eusterholza,d, Martin Heilmaiera and Alexander Kauffmann*e a Institute for Applied Materials (IAM-WK), Karlsruhe Institute of Technology (KIT), Kaiserstraße 12, 76131 Karlsruhe, Germany b Department of Materials Science and Engineering, Kyoto University, Sakyo-kyu, Kyoto 606-8501, Japan c Laboratory for Electron Microscopy (LEM), Karlsruhe Institute of Technology (KIT), Engesserstr. 7, 76131 Karlsruhe, Germany d Karlsruhe Nano Micro Facility (KNMFi), Karlsruhe Institute of Technology (KIT), Hermann-vonHelmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany e Institute for Materials (IM), Ruhr University Bochum (RUB), Universitätsstraße 150, 44780 Bochum, Germany * corresponding author mail: [email protected] (A. Kauffmann) phone: +49 234 32 18430 Abstract Recently, a new generation of high strength steels was introduced by utilizing a lateral chemical pattern 1 of an austenite stabilizer to create microstructures of austenite γ and martensite α’ after quenching. A 2 ternary Fe-Mn-C pearlite is a suitable initial state for this if Mn effectively partitions into the cementite. 3 In the present study, two model Fe-Mn-C alloys were pearlite treated outside the well-established local 4 equilibrium, Mn partitioning regime (P-LE). A complete pearlite formation was achieved not only for 5 Fe-3.0Mn-3.0C (at.%, Alloy A) at high pearlite formation temperature but also for Fe-6.9Mn-3.2C 6 (at.%, Alloy B) at low transformation temperature. The morphology of the pearlite included fine-scaled 7 fibers and lamellae. Even though outside the P-LE region, significant Mn partitioning into cementite 8 was obtained for both alloys. Pearlite was formed at approximately the overall Mn content, while 9 growing either enriched or depleted in C for most of the reaction. The successful application of a short 10 time austenitization treatment was proven for both alloys transforming the pearlite into α’ + γ 11 microstructures while retaining the initial pearlite morphology. Thus, fine-structured α’ + γ can be 12 synthesized from pearlite processed well outside the established Mn partitioning regimes, opening a 13 much larger compositional and processing space. 14 Keywords Pearlite, Austenite, Martensite, Fe-Mn-C, Transformation 15 2 1 Introduction In 2018, Sun et al. [1] introduced a promising processing route to achieve an outstanding combination 16 of strength and ductility in Mn containing steels by fine-scaled austenite-martensite microstructures. 17 The basic premise is to form a chemical Mn pattern in austenite γ-(Fe,Mn,C) prior to quenching. The 18 Mn pattern gets introduced via an initial pearlite formation, where Mn is strongly enriched in the 19 cementite θ-(Fe,Mn)3C compared to the ferrite α-(Fe,Mn,C). These phases will be abbreviated as γ, θ 20 and α, respectively. In a subsequent short time austenitization (STA) treatment, the chemical Mn pattern 21 remains stable under suitable processing conditions, even after the complete transformation to γ. Rapid 22 cooling after completion of the γ transformation results in a microstructure similar in morphology and 23 dimensions to the initial pearlite, consisting of martensite α’-(Fe,Mn,C) (Mn depleted regions, 24 abbreviated with α’) and austenite γ (Mn enriched regions). By additional tempering treatments, 25 remarkable ultimate tensile strength and strain to failure combinations of 1.6-2.1 GPa and 7-10 % were 26 achieved [1]. 27 These results were achieved with a hypo-eutectoid alloy composition of Fe-4.3Mn-2.3C (at.%, Fe28 4.4Mn-0.5C in wt.%) [1]. To test the limits of this scheme, one obvious possibility would be the 29 manipulation of the alloy composition. As the final microstructure and therefore also the resulting 30 mechanical properties strongly depend on the Mn pattern, a variation of the overall Mn content would 31 be a feasible objective for tailoring the mechanical properties. However, the recent endeavors on this 32 topic [1–5], including the original work, strictly focus on Mn contents in a range of only 2-5 at.% Mn 33 and a C content that results in an overall hypo-eutectoid composition. This narrow compositional range 34 might be rationalized by what was suggested in the early work of Hutchinson et al. [6] on the growth 35 kinetics and local equilibrium (LE) conditions of Fe-Mn-C pearlite. Their work presents two different 36 design principles to predict the LE at the reaction front for pearlite formation: one for transformations 37 occurring in the three-phase region of α + θ + γ, and another for the two-phase region of α + θ. 38 The design principles in Ref. [6] are strictly limited by the construction of two Mn partitioning 39 boundaries that originate from a separate LE design approach for pro-eutectoid α and θ formation. The 40 approach is based on the vastly different diffusion coefficients of interstitial, fast-diffusing C compared 41 to substitutional, slow-diffusing Mn. The diffusivity ratio is about 104 – 106 [7]. The boundaries mark 42 the compositional and temperature transition to a reaction where Mn partitioning is no longer 43 thermodynamically necessary between α/γ and θ/γ, respectively. These fundamentals have first been 44 transferred qualitatively from the pro-eutectoid reactions to pearlite formation by Coates and Hillert [7, 45 8], as the corresponding reaction front exhibits LEs for both α/γ and θ/γ. With the simultaneous 46 application of both partitioning boundaries, two major regimes were derived: (i) a partitioning regime 47 (P-LE) where the partitioning criteria with γ are fulfilled for both α and θ and (ii) a negligible48 partitioning regime (NP-LE) where the criteria are not satisfied for either of the two. As the pearlite 49 formation with a sufficiently strong Mn partitioning is a prerequisite for the aforementioned processing 50 route, the introduction of these regimes heavily restricts the applicable range of alloy compositions and 51 transformation temperatures, as staying inside the P-LE regime seemed reasonable [6]. 52 What has not been experimentally addressed in literature thus far is the existence of an additional third 53 and fourth regime besides P-LE and NP-LE. In these regimes, only one of the two partitioning criteria 54 is fulfilled, either for α or θ. The regime of interest for this work is the one that fulfills the Mn partitioning 55 criterion for α, but not for θ. It includes a large range of temperatures and compositions, especially at 56 higher Mn and C contents, which have yet to be tested for Mn partitioning and the possibility to apply 57 the intended STA processing. However, with increasing Mn content the eutectoid line as well as the 58 3 three-phase field that separates the γ single-phase field and the α + θ two-phase field, shift to lower 59 temperatures [9]. A lower transformation temperature as well as high contents of the slow-diffusing Mn 60 will result in a currently unknown retardation of the pearlite transformation. Another aspect that needs 61 to be considered at such retarded reaction velocities is the formation of metastable phases or other 62 microstructures besides α + θ pearlite. In case of Fe-Mn-C, potential formation of M5C2 [10] and M23C6 63 carbides needs to be considered, with M representing metallic elements. Their formation has been 64 reported for example in Cr-rich and Al-rich, Mn-containing steels [10, 11]. 65 Thus, the following research questions will be addressed in the present study: 66 1. Does a pearlite treatment for a low Mn containing Fe-3.0Mn-3.0C (at.%) at 600 °C and for a 67 high Mn containing Fe-6.9Mn-3.2C (at.%) at 540 °C result in an entirely transformed 68 microstructure comprising pearlite that only consists of α and θ? 69 2. Does pearlite in these Fe-Mn-C alloys exhibit considerable Mn partitioning into θ and how does 70 it compare to global equilibrium (GE) and LE conditions? 71 3. Is it possible to achieve a α’ + γ microstructure by short-time austenitization from these pearlite 72 conditions? 73 2 Experiments and Simulations The design of the Mn partitioning boundaries was first developed for pro-eutectoid α and θ formation [7, 74 12] and quantitatively applied to Fe-Mn-C pearlite formation by Hutchinson et al. [6]. Their application 75 for given isothermal sections is shown in Fig. 1. It divides the parameter space, described by 𝑋C and 76 𝑋Mn as the molar fraction of C and Mn, respectively, into four regions. The representation in Fig. 1, 77 however is given by using the derived quantities 𝑈C=𝑋C/(1−𝑋C) and 𝑈Mn =𝑋Mn/(1−𝑋C) 78 according to the treatment by Hillert [8, 12] as well as Coates [7] and Hutchinson [6]. A detailed 79 explanation of the physical meaning and practical usage of these variables can be found in the 80 Supplementary Material of this article. The temperatures of 600 (Fig. 1 a)) and 540 °C (Fig. 1 b)) were 81 selected to capture a potentially fast transformation at low Mn contents and a retarded one at high Mn 82 contents, respectively. The design of the partitioning boundaries (dot-dashed, purple lines in Fig. 1) are 83 adopted from Ref. [7] and comprehensively described for the present cases in the Supplementary 84 Material, Fig. S1. The thermodynamic data was obtained using the 2023 PanHEA database in the Pandat 85 software provided by CompuTherm (USA). The theoretical data on the phase field positions of the Fe86 Mn-C system are in good agreement with experimental literature data in the relevant compositional 87 range [1, 9, 13–21]. All additional thermodynamic LE considerations that are not captured by Pandat 88 were obtained using self-developed Matlab scripts which are published public domain [22]. 89 As mentioned in Sec. 1, out of the four regimes formed by the use of the two partitioning boundaries, 90 only two have been addressed in literature thus far [6–8]: The P-LE regime (highlighted in purple) and 91 the NP-LE regime (highlighted in gray). For the pearlite transformation, these two regimes are uniquely 92 defined by the fulfillment of the two independent Mn partitioning criteria between α and γ as well as θ 93 and γ (see Supplementary Material Fig. S1). 94 The regime to the right of the intersection of the two dot-dashed purple lines in Figs. 1 a) and b) is the 95 scope of the present article. It includes high overall Mn contents which hold the potential of stabilizing 96 large fractions of γ in the final microstructure after STA. Here, the partitioning criteria are fulfilled for 97 α, but not for θ formation. 98 Inside this regime two model alloys, Alloy A and Alloy B, were selected. They lie inside the α + θ two99 phase field as well as inside the metastable extensions of the α + γ and γ + θ phase field boundaries to 100 4 avoid pro-eutectoid α or θ formation. The nominal compositions of the Mn-lean Alloy A and the Mn101 rich Alloy B are Fe-3.0Mn-3.0C and Fe-6.9Mn-3.2C (at.%, Fe-3.0Mn-0.7C and Fe-7.0Mn-0.7C in 102 wt.%), respectively. 103 Fig. 1: Fe-Mn-C isothermal sections at a) 600 °C and b) 540 °C. Black bold lines illustrate the phase fields in GE. Black dotted lines indicate the metastable extrapolations of the α + γ and γ + θ phase field boundaries. The purple dot-dashed lines represent Mn partitioning boundaries for α (near horizontal) and θ (near vertical). P-LE and NP-LE regions are marked purple and gray, respectively. Alloys A and B, both lie outside these regions. The experimentally determined compositions listed in Tabs. 1 and 2 are displayed. Alloy manufacturing was done by arc melting high purity bulk elements as well as an in-house 104 synthesized Fe3C obtained by the same method. The elements Mn (etched nominal purity 99.8 %) and 105 graphite (nominal purity 99.999 %) were provided by chemPUR GmbH (Germany). Fe (nominal purity 106 99.99 %) was provided by Alfa Aesar (United States). Arc melting was conducted in an AM/0.5 furnace 107 provided by Edmund Bühler GmbH (Germany). The furnace chamber was evacuated to 5∙10−2 mbar 108 and filled with Ar. This process was repeated for three times in total in order to purify the melting 109 atmosphere. Then, a vacuum of less than 1∙10−4 mbar was established. The processing chamber was 110 then filled with Ar once more to a pressure of 600 mbar. Residual O2 within the furnace chamber was 111 gettered by liquefying a Zr granule before melting the bulk elements. Every manufactured ingot was 112 flipped and re-melted at least five times. The alloy ingots were homogenized utilizing a STF15/450 21113 601449 tube furnace by Carbolite Gero GmbH & Co. KG (Germany) with flowing high purity Ar 114 atmosphere. The heat treatment temperature was 1100 °C for a dwell time of 96 h. Heating and cooling 115 were conducted at a rate of 115 K/h. The chemical composition was determined through optical emission 116 spectroscopy (OES) by analyzing both, the top and bottom side after the last re-melting step. The results 117 are displayed in Tabs. 1 and 2 for Alloys A and B, respectively. The standard deviation for all elements 118 was determined to be < 0.07 at.% (0.07 wt.%). Samples for further investigations were cut from the 119 alloy batches using a high-speed rotatory cutting device (Struers, France). 120 Pearlite formation was carried out in air at atmospheric pressure using two pre-heated box furnaces. 121 After completion of the prior austenitization step at 910 °C for 1 h, an immediate furnace transfer was 122 performed. Pearlite in Alloy A and Alloy B was formed at 600 °C for 16 h and 540 °C for 96 h, 123 respectively, based on initial trials to obtain relevant temperature/time combinations. The heat 124 treatments were concluded by oil quenching. 125 The STA treatments were done in a pre-heated box furnace at 770 °C for 150 s and also concluded by 126 oil quenching. To ensure a high heating rate, the samples were covered by pre-heated Al2O3 powder at 127 the start of the treatment. 128 5 The samples for microstructure characterization were prepared by water-cooled grinding with SiC paper 129 followed by standard metallographic polishing steps with 3 and 1 µm polycrystalline diamond 130 suspensions. As a finishing step, polishing with MasterMet-2 by Buehler (USA) was performed for 131 5 min. Samples were subsequently etched in a 1 % Nital solution for 5 s. The grinding process was 132 carried out to remove any decarburized layer that has formed during the heat treatments. Microstructure 133 characterization was performed by scanning electron microscopy (SEM). For secondary electron (SEM134 SE) as well as backscattered electron (SEM-BSE) contrast imaging, a Leo Gemini 1530 field-emission 135 SEM by Zeiss (Germany) was utilized. Low magnification, wide field of view micrographs with high 136 resolution were obtained by automated stitching in Microsoft Image Composite Editor. The 137 quantification of microstructural details including the area fractions of the constituting phases and 138 morphologies was carried out using the ImageJ software. Estimates for the fractions of regions of similar 139 morphology were done by manually selecting and asserting regions with θ aspect ratios close to one as 140 fibrous. Fibers oriented parallel to the sample surface were therefore identified as lamellae. Micrographs 141 for all analysis were selected randomly and at different magnification to enable both a large overall area 142 of investigation as well as high accuracy of the imaging. The difference in total area mapped between 143 different magnification micrographs was considered. The determined area fractions were converted into 144 volume fractions under the assumption of isometry and isotropy. A more detailed assessment of 145 parameters like θ spacing or θ morphology fraction would require 3D methods for microstructure 146 analysis, e.g. FIB tomography, and was not scope of the present study. 147 Phase identification was done by means of transmission synchrotron X-ray diffraction (XRD) at 148 beamline BL13XU, SPring-8 of the Japan Synchrotron Radiation Research Institute (JASRI). The 149 diffraction experiments were performed using monochromatic synchrotron radiation at an energy of 150 30 keV ( λ = 0.0413 nm) and 30 s exposure time. The calibration of the 2θ zero-shift and sample 151 displacement was done with a standard specimen of ceria (CeO₂). Additionally, local analysis of the 152 constituting phases was performed via selected area electron diffraction (TEM-SAED) with a JEM153 2100F by Jeol (Japan) and a Titan3 by FEI (USA) transmission electron microscope (TEM) operated at 154 200 kV and 300 kV, respectively. For determining the local composition of the phases, both energy 155 dispersive X-ray spectroscopy in scanning TEM (STEM-EDS) as well as atom probe tomography (APT) 156 were considered. However, due to significant inaccuracies introduced by quantifying the C content, APT 157 was selected as the superior method both in terms of data accuracy and spatial resolution [23]. The 158 experiments were performed on both alloys after pearlite formation with a LEAP 4000X HR by Cameca 159 (France). Data presented in the main article were generated in voltage mode, while additional data 160 depicted in the Supplementary Material were generated in laser mode. Tip preparation was done via 161 focused ion beam (FIB) milling with an Auriga 60 by Zeiss (Germany). Tips presented in the 162 Supplementary Material were coated with Cr to increase tip stability, following the routine used in 163 Ref. [24]. Peak decomposition was required for the θ regions, which was applied by the integrated 164 feature of the AP Suite 6.3 software (Cameca, France). The software decomposes overlapping peaks by 165 considering the remaining unique peak(s) of the species of interest of the same charge and applies the 166 relative natural abundances to weigh the contribution of that species to the initial peak. An example for 167 this would be C2 1+ and C4 2+ in θ. Both species exhibit peaks at 24 and 25 Da. However, C4 2+ possesses a 168 unique peak at 24.5 Da as well. Considering the natural abundances of the different C4 2+ isotopes, the 169 contribution to the peak at 24 Da and 25 Da can be derived, which leaves the remaining counts of the 170 peaks for C2 1+. 171 6 3 Results and Discussion 3.1 Pearlite Formation To answer the research questions from Sec. 1, it is essential to develop an understanding of the 172 intricacies of a complete Fe-Mn-C pearlite formation and the Mn partitioning behavior. Hillert [12] 173 proposed that for simultaneous Mn and C partitioning, the driving force for C diffusion in γ ahead of the 174 reaction front must be reduced to compensate for the orders of magnitude difference in diffusion 175 coefficients of the elements. A practical way to achieve this condition was shown by Hutchinson 176 et al. [6], i.e. γ in contact with either growing θ or α are required to share the same C activity aC γ. This 177 can be elegantly applied quantitatively to isothermal sections of the ternary phase diagram, as depicted 178 in Fig. 2. The compositions of γ ahead of growing θ and α are located on the same C iso-activity line in 179 Fig. 2 (blue dots on the C iso-activity line at the intersections with the dotted, metastable extensions). 180 The forming θ and α are obtained from following the tie lines (blue, solid lines) connected to these γ 181 compositions. As Hutchinson et al. [6] obtained a steady-state growth of pearlite within the α + θ two182 phase field with constant growth rate and interphase spacing, an additional boundary condition regarding 183 the overall composition of the growing pearlite was introduced. Namely, its composition must be 184 consistent with the alloy composition. This can only be achieved when the connecting line (blue, dashed 185 line) between the growing θ and α intersects the alloy composition. This connecting line does not 186 represent a tie line as the two growing phases are not in equilibrium with each other. In what follows, 187 this design principle is referred to as steady-state, C activity-based local equilibrium condition (SCA188 LE). As depicted in Fig. 2, the application of the SCA-LE design indicates an enrichment of Mn in θ 189 during pearlite formation even though it is being applied to Alloys A and B outside the P-LE regime. 190 Fig. 2: Schematic design according to SCA-LE in isothermal sections: a) Alloy A at 600 °C and b) Alloy B at 540 °C. The illustration of the isothermal section is the same as for Fig. 1. The blue solid lines indicate tie lines while the blue dashed line is the line connecting θ and α forming during steady-state pearlite growth. The predicted enrichment of Mn in θ is: a) 8.0 at.% (vs. 2.8 at.% in Alloy A) and b) 35.0 at.% (vs. 6.9 at.% in Alloy B). In Fig. 3, micrographs of both alloys are shown after their respective pearlite treatments: 600 °C/16 h 191 for Alloy A and 540 °C/96 h for Alloy B. A complete transformation was achieved for both alloys. High 192 resolution, wide field of view micrographs obtained from stitching proving the statement are provided 193 via a public repository [25]. Despite the lower transformation temperature, which is usually associated 194 with a decrease in lamellar spacing [26], Alloy B does not show a clearly refined microstructure 195 compared to Alloy A. With respect to pearlite morphology, a notable difference between the two alloys 196 is obtained. Both exhibit extended regions of fibrous morphology, comprising a volume fraction of 197 (0.91 ± 0.08) and (0.32 ± 0.12) of the pearlite in Alloys A and B, respectively. The volume-specific 198 phase fraction of θ is (0.06 ± 0.01) for Alloy A and (0.21 ± 0.02) for Alloy B. Alloy B frequently 199 7 exhibits regions of increasing interface spacing towards the colony boundaries, thus, at potentially late 200 stages of the transformation before completion. Due to their low overall fraction of (0.02 ± 0.01), these 201 regions were not included in the randomly selected micrographs to determine the volume-specific phase 202 fractions. 203 Fig. 3: SEM-SE micrographs (Nital etching) after completion of pearlite formation: a) Alloy A, 600 °C/16 h and b) Alloy B, 540 °C/96 h. No untransformed regions were found. θ appears light, α appears dark. Lamellar morphology is indicated by yellow, fibrous by red arrows. High resolution, wide field of view micrographs obtained from stitching are available via Ref. [25]. To identify the constituting phases of both alloys after the pearlite treatment, transmission XRD 204 experiments were performed. Due to large pearlite colony size, preferred orientation and intensities 205 different from powder diffraction patterns were obtained. All analyzed peaks were exclusively attributed 206 to either α or θ, as shown in Fig. 4 and no evidence for any other carbide than θ or untransformed γ was 207 found. The (211)θ and (112)θ peaks were considered to uniquely identify the carbide as θ. 208 Fig. 4: Transmission XRD patterns of: a) Alloy A and b) Alloy B. The intensity is plotted as a function of 2Θ. Characteristic diffraction peaks corresponding to α are marked in red, while peaks associated with θ are highlighted in blue. All relevant peaks of θ are highlighted below the spectra, while the (211)θ and (112)θ peaks were used to uniquely identify the carbide Further characterization of the pearlite was done via TEM imaging and electron diffraction. Due to the 209 large pearlite colony size, TEM specimens were taken from representative regions, i.e. a fibrous colony 210 for Alloy A and a lamellar colony for Alloy B. Both lift outs were taken in regions of near constant 211 interface spacing (73 ± 26 nm) and (104 ± 19 nm), respectively. Bright field micrographs as well as the 212 TEM-SAED patterns are shown in Fig. 4. The zone axes in both Fig. 4 a) and d) are parallel to the 213 [100]θ direction following the lattice parameter convention, 𝑏 > 𝑎 > 𝑐 for θ. The lift outs were 214 chosen perpendicular to the long fiber axes for Alloy A (Fig. 4 a)) and the lamellar plane for Alloy B 215 8 (Fig. 4 c)). In such a scenario, fibrous and lamellar colonies of interest cannot be distinguished from one 216 another as both appear similar at the sample surface. They can only be distinguished after cutting a FIB 217 trench and subsequent cross-sectional imaging. For both samples, the transmission XRD results 218 regarding the constituting α and θ phases were confirmed by TEM-SAED patterns in Figs. 4 b) and d), 219 respectively. Furthermore, the orientation relationship (OR) was determined. The investigated colonies 220 of Alloy A and Alloy B both exhibit variants of the Isaichev OR with 〈111〉α || 〈100〉θ and 221 {12 1}α || {011}θ, which is well established for pearlite [27]. Please note that Fig. 4 a) for Alloy A 222 exhibits a twin site of the OR with a (011 )θ habit plane instead of (011)θ as for Fig. 4 b). The θ twin 223 variant is of compound character with the habit plane (011)θ, shear direction [011 ]θ and shear plane 224 (100)θ. Furthermore, the twin orientation can be achieved via a rotation of the θ matrix crystal around 225 the [100]θ direction by 111.8°. This equates to the angle between (011)θ and (011 )θ in the θ unit cell. 226 The Isaichev OR might alternatively be expressed differently by defining the habit plane according to 227 {01 1}α || {031}θ. This can be attributed to the very small angular mismatch (0.3°) between the sets of 228 lattice planes involved. However, the angular mismatch becomes more apparent for high indexed spots. 229 Fig. 5: TEM bright field micrographs (a),c)) and corresponding TEM-SAED patterns (b), d)) of: a), b) fibrous regions from Alloy A and c),d) lamellar region from Alloy B. The zone axes (ZA) in both b) and d) are [111]α || [100]θ. Samples for APT were selected in similar fashion like the TEM specimens, i.e., lift out via FIB in a 230 fibrous colony for Alloy A and in a lamellar colony for Alloy B, both within regions of nearly constant 231 interface spacing. In total, four tips were analyzed with two for Alloy A stemming from the same colony 232 and two for Alloy B from different colonies. One of each alloy is discussed as an example. The 233 remaining data agrees with the statements and discussion. The data is included in Fig. S2, Fig. S3 and 234 Tab. S2 of the Supplementary Material. To allow for a more detailed assessment of the chemical 235 composition, the local phase fractions were determined at the lift out position from SEM-SE 236 micrographs. The local θ volume fraction of Alloy A is 0.05, while it is 0.19 for Alloy B. These values 237 were then converted to the local molar phase fractions 𝑓m which are provided together with the values 238 for α in Tabs. 1 and 2 for Alloys A and B, respectively. 239 9 Tab. 1: Compositions 𝑋, C activity in γ a C γ, molar phase fractions 𝑓m and 𝑈 of Alloy A according to thermodynamic calculations (GE, SCA-LE), experimental (exp.) results from APT analyses and reconstructions. + denotes experimental data. # denotes experimental data adjusted according to the thermodynamic dataset. The other data was derived from the thermodynamic dataset. Condition Phase Position 𝑋C 𝑋Mn 𝑈C 𝑈Mn a C γ 𝑓m Alloy A – – 0.029 + 0.028 + 0.030 0.029 – – GE α – 1 · 10-4 0.011 1 · 10-4 0.011 – 0.88 θ – 0.250 0.163 0.333 0.217 – 0.12 SCA-LE α γ 1 · 10-4 0.021 1 · 10-4 0.021 – 0.88 θ γ 0.250 0.080 0.333 0.106 – 0.12 γ α 0.031 0.114 0.032 0.118 0.29 – θ 0.020 0.027 0.020 0.027 0.29 – Exp. α – 1 · 10-4 + 0.020 + 1 · 10-4 0.020 – 0.95 + θ – 0.262 + 0.218 + 0.355 0.295 – 0.05 + Reconstr. α γ 1 · 10-4 + 0.020 + 1 · 10-4 0.020 – 0.95 + θ γ 0.250 # 0.222 # 0.333 0.296 – 0.05 + γ α 0.034 0.111 0.035 0.115 0.33 – θ 0.017 0.089 0.017 0.091 0.16 – Tab. 2: Compositions 𝑋, C activity in γ a C γ, molar phase fractions 𝑓m and 𝑈 of Alloy B according to thermodynamic calculations (GE, SCA-LE), experimental (exp.) results from APT analyses and reconstructions. + denotes experimental data. # denotes experimental data adjusted according to the thermodynamic dataset. The other data was derived from the thermodynamic dataset. Condition Phase Position XC XMn UC UMn a C γ 𝑓m Alloy B – 0.029 + 0.069 + 0.030 0.071 – – GE α – 1 · 10-5 0.029 1 · 10-5 0.029 – 0.88 θ – 0.250 0.375 0.333 0.500 – 0.12 SCA-LE α γ 1 · 10-5 0.032 1 · 10-5 0.032 – 0.88 θ γ 0.250 0.350 0.333 0.466 – 0.12 γ α 0.013 0.179 0.013 0.181 0.09 – θ 0.010 0.147 0.010 0.148 0.09 Exp. α – 2 · 10-4 + 0.020 + 2 · 10-4 0.020 – 0.80+ θ – 0.221 + 0.293 + 0.284 0.376 – 0.20 + Reconstr. α γ 1 · 10-4 # 0.020 + 1 · 10-4 0.020 – 0.80 + θ γ 0.250 # 0.282 # 0.333 0.376 – 0.20 + γ α 0.054 0.158 0.058 0.167 0.76 – θ 0.011 0.106 0.011 0.107 0.14 – Two reconstructed tips are depicted in Fig. 6. Both show Mn and C enrichment coinciding in the same 240 region, a θ fiber (Fig. 6 a)) for Alloy A and lamella (Fig. 6 b)) for Alloy B. The determined C content 241 of θ in all tips (Tab. 1, Tab. 2 and Tab. S2 in the Supplementary Material) deviates from the 242 stoichiometric expectation, which may be explained by different phenomena. Possible scenarios are 243 discussed in Ref. [28, 29]. An additional aspect that could have affected the evaporation and, therefore, 244 16 [8] Hillert, M.: Phase Equilibria, Phase Diagrams and Phase Transformations: Their Thermodynamic 446 Basis. In: Phase Equilibria, Phase Diagrams and Phase Transformations: Their Thermodynamic 447 Basis. Cambridge: Cambridge University Press, 2007, p. 358–366. 448 [9] Benz, R. et al.: Thermodynamics of the solid phases in the system Fe−Mn−C. Metall Trans, 4 (8), 449 1973, p. 1975–1986. 450 [10] Shen, Y.Z. et al.: M5C2 Carbides in a High-Chromium Ferritic/Martensitic Steel. Metall Mater 451 Trans A, 45 (7), 2014, p. 2950–2962. 452 [11] Cheng, W.-C. et al.: On the Eutectoid Reaction in a Quaternary Fe-C-Mn-Al Alloy: 453 Austenite → Ferrite + Kappa-Carbide + M23C6 Carbide. Metall Mater Trans A, 45 (3), 2014, p. 454 1199–1216. 455 [12] Hillert, M.: Swedish Institute for Metal Research, 1953. 456 [13] Eckel, J.F., Krivobok, V.N.: Alloys of iron, manganese and carbon-part VIII. Trans. Am. Soc. St. 457 Tr., 21, 1933, p. 846–864. 458 [14] Gensamer, M.: Alloys of iron and manganese-part XII. Trans. Am. Soc. St. Tr., 21, 1933, p. 1028– 459 1034. 460 [15] Hillert, M., Waldenström, M.: A thermodynamic analysis of the Fe-Mn-C system. Metall Trans A, 461 8 (1), 1977, p. 5–13. 462 [16] Stefanescu, D., Katz, S.: Thermodynamic Properties of Iron-Base Alloys. In: Casting (Editors: S. 463 Viswanathan et al.). Ohio: ASM International, 2008, p. 41–55. 464 [17] Bain, E.C. et al.: The equilibrium diagram of iron-manganese-carbon alloys of commercial purity. 465 Tran. Am. Inst. Min. Metall. Eng., (100), 1932, p. 228–256. 466 [18] Tofaute, W., Linden, K.: Die Umwandlung im festen Zustande bei Manganstählen mit Gehalten bis 467 1,2% C und 17% Mn. Archiv für das Eisenhüttenwesen, 10 (11), 1937, p. 515–524. 468 [19] Vogel, R., Döring, W.: Das System Eisen-Zementit-Mangankarbid-Mangan. Archiv für das 469 Eisenhüttenwesen, 9 (5), 1935, p. 247–252. 470 [20] Walters, F.M., Wells, C.: Alloys of iron and manganese-part XIII. Trans. ASM, 23 (3), 1935, p. 471 727–750. 472 [21] Walters, F.M., Wells, C.: Alloys of iron, manganese and carbon-part XIV. Trans. ASM, 23 (3), 473 1935, p. 751–760. 474 [22] Muench, M., Kauffmann, A.: Matlab code to ‘Assessment of the compositional requirements to 475 form Fe-Mn-C austenite-martensite composites’. Karlsruhe Institute of Technology, 2025. 476 [23] Gault, B. et al.: Atom probe tomography. Nat Rev Methods Primers, 1 (1), 2021, p. 1–30. 477 [24] Schwarz, T.M. et al.: In Situ Metallic Coating of Atom Probe Specimen for Enhanced Yield, 478 Performance, and Increased Field-of-View. Microscopy and Microanalysis, 30 (6), 2024, p. 1109– 479 1123. 480 [25] Muench, M., Kauffmann, A.: Completion of Pearlite Transformation in Fe-Mn-C Alloys with 3 and 481 7 wt.% Mn. Zenodo, 2025. 482 [26] Mehl, R.F., Hagel, W.C.: The austenite: Pearlite reaction. Progress in Metal Physics, 6, 1956, p. 483 74–134. 484 17 [27] Zhou, D.S., Shiflet, G.J.: Ferrite: Cementite crystallography in pearlite. Metall Trans A, 23 (4), 485 1992, p. 1259–1269. 486 [28] Peng, Z. et al.: Unraveling the Metastability of Cn2+ (n = 2–4) Clusters. J. Phys. Chem. Lett., 10 487 (3), 2019, p. 581–588. 488 [29] Takahashi, J. et al.: Study on Quantitative Analysis of Carbon and Nitrogen in Stoichiometric θ489 Fe3C and γ′-Fe4N by Atom Probe Tomography. Microscopy and Microanalysis, 26 (2), 2020, p. 490 185–193. 491 492