In Situ Observations of the Microstructural Evolution during Heat Treatment of a PH 13-8 Mo Maraging Steel

The standard heat treatment of PH 13-8 Mo maraging steels consists of solution annealing and subsequent aging. Herein, it is investigated how an additional intercritical annealing step prior to aging affects the microstructure, and, consequently, the mechanical properties of a PH 13-8 Mo maraging steel. In situ techniques by means of high-temperature electron backscatter diffraction and high-temperature X-ray diffraction are applied to study the microstructural changes during intercritical annealing and subsequent aging. In addition, high-resolution investigation methods, such as transmission electron microscopy and atom probe tomography supplemented by transmission Kikuchi diffraction, are used for an in-depth characterization of the microstructure. The results reveal that a diffusion-controlled martensite to austenite transformation accompanied by partitioning of the substitutional atoms Cr, Ni, and Mo takes place during intercritical annealing. As a result of partitioning during intercritical annealing, an inhomogeneous distribution of Ni remains in the microstructure after the martensitic transformation. Consequently, the formation of reverted austenite is facilitated during subsequent aging due to existing Ni-enriched zones in martensite. Since the fracture toughness is signi ﬁ cantly enhanced compared to the standard heat treatment, it is suggested that this improvement is related to the increased phase fraction of reverted austenite.

The standard heat treatment of PH 13-8 Mo maraging steels consists of solution annealing and subsequent aging. Herein, it is investigated how an additional intercritical annealing step prior to aging affects the microstructure, and, consequently, the mechanical properties of a PH 13-8 Mo maraging steel. In situ techniques by means of high-temperature electron backscatter diffraction and high-temperature X-ray diffraction are applied to study the microstructural changes during intercritical annealing and subsequent aging. In addition, highresolution investigation methods, such as transmission electron microscopy and atom probe tomography supplemented by transmission Kikuchi diffraction, are used for an in-depth characterization of the microstructure. The results reveal that a diffusion-controlled martensite to austenite transformation accompanied by partitioning of the substitutional atoms Cr, Ni, and Mo takes place during intercritical annealing. As a result of partitioning during intercritical annealing, an inhomogeneous distribution of Ni remains in the microstructure after the martensitic transformation. Consequently, the formation of reverted austenite is facilitated during subsequent aging due to existing Ni-enriched zones in martensite. Since the fracture toughness is significantly enhanced compared to the standard heat treatment, it is suggested that this improvement is related to the increased phase fraction of reverted austenite.
510°C for 4 h in ref. [10]. In general, higher phase fractions of retained austenite (i.e., remaining austenite after full or partial austenitization) are anticipated in martensitic steels after intercritical annealing since the partial reversion of martensite enforces a redistribution of austenite-stabilizing elements, such as Ni or Mn. This enrichment in austenite suppresses the martensitic transformation upon cooling, leading to a considerable amount of retained austenite at room temperature. [15][16][17][18][19][20][21][22][23][24][25][26] The higher the temperature during intercritical annealing, the less stabilizing elements are required for austenite to be in the thermodynamic equilibrium. [21] Hence, the martensitic transformation is facilitated and less austenite is retained at ambient temperatures. As reported in the study by Shirazi et al. [27], the dependence of the phase fraction of retained austenite on the temperature during intercritical annealing can be also attributed to the evolving grain size of this phase. Smaller grains resulting from low annealing temperatures may restrict a multivariant martensitic transformation, thus, a considerably larger chemical driving force for single-variant transformation is required. [27,28] In summary, several studies reported a significant increase in the phase fraction of austenite, either retained or reverted, for various martensitic steels when intercritical annealing had been integrated. Consequently, improved toughness and ductility were obtained. However, there is a lack of knowledge about the microstructural changes during intercritical annealing of PH 13-8 Mo maraging steels, particularly whether this additional annealing step promotes the formation of reverted austenite at higher aging temperatures. Apart from the microstructural changes themselves, the influence of the final microstructure on the mechanical properties, particularly on the toughness, requires elaborate investigation and discussion. This study aims to answer these questions by combining in situ techniques by means of high-temperature electron backscatter diffraction (HT-EBSD) and high-temperature X-ray diffraction (HT-XRD) with ex situ experiments, such as transmission electron microscopy (TEM), transmission Kikuchi diffraction (TKD), and atom probe tomography (APT) for a thorough correlative investigation. Ultimately, mechanical testing by means of Vickers hardness testing and fracture toughness determination was conducted to assess the effect of the microstructural changes resulting from intercritical annealing on the mechanical properties of PH 13-8 Mo maraging steels.

Material and Heat Treatment
The chemical composition of the investigated material is displayed in Table 1. The alloy was provided in the form of hot-forged round bars with 180 mm in diameter by voestalpine BÖHLER Aerospace GmbH & Co KG. Dilatometry samples with 5 and 8 mm in diameter were machined from half the radius of the round bar. A variation of heat treatments listed in Table 2 was conducted in a DIL 805A dilatometer from TA instruments. The focus was laid on varying the intercritical annealing temperature and a comparison with the standard heat treatment (SA). In addition, an extended heat treatment comprising of a double sequence of solution and intercritical annealing prior to aging was performed (SI 760 SI 760 A). While each heating step was operated in vacuum atmosphere, N 2 gas was employed during isothermal holding and subsequent cooling for an improved temperature control. Figure 1 shows the dilatation curve of the investigated steel for continuous heating with 1.5 K s À1 after solution annealing. The graph demonstrates that for this given heating rate, temperatures of 700, 730, and 760°C are situated between A s and the austenite finish temperature (A f ), corresponding to intercritical  annealing in the dual phase field of martensite and austenite. It is noteworthy that after solution and intercritical annealing the samples were further cooled below 16°C in ice water to eliminate excessive amounts of retained austenite.

Investigation Methods
After the heat treatments, the dilatometry samples were mechanically grounded and polished, using 1 μm diamond suspension in the final step, for a microstructural characterization of reverted austenite. For EBSD measurements, the samples underwent electrolytic polishing at room temperature with Struers A2 electrolyte for 20 s in a Struers LectroPol-5 operating at 20 V. A FEI Versa 3D DualBeam workstation equipped with an EDAX Hikari EBSD system was employed for the EBSD scans. The acceleration voltage was set to 20 kV and a step size of 70 nm was used. Subsequently, the EBSD data were analyzed with the software OIM Analysis v8.6 from EDAX. A grain dilatation cleanup with an exclusion of points exhibiting a confidence index (CI) lower than 0.2 was applied to the EBSD data to reduce noise. Moreover, the grain tolerance angle was set to 5°and the minimum grain size to 2 points. To determine the volume phase fraction of austenite, XRD diffractograms were captured on an 1D detector of a Bruker-AXS D8 Advance DaVinci diffractometer with Bragg-Brentano geometry in the 40-100°2θ angular range. Cu K-α radiation was used and 40 kV was selected for the X-Ray source, while the current was adjusted to 40 mA. The diffractograms were evaluated via Rietveld analysis using the software TOPAS v6 from Bruker-AXS. The Pseudo-Voigt function was selected for peak fitting and four peaks each for martensite and austenite were considered during Rietveld refinement. Since PH 13-8 Mo maraging steels exhibit low lattice distortion from interstitial elements, [4] the ferrite structure file was used for fitting the martensite peaks. For each determined phase fraction, an error was calculated by the software. The magnitude of the error is mainly dependent on the goodness of fit during the Rietveld refinement, and the number of instrumental and microstructural parameters which are set to be refined. Vickers hardness measurements were carried out according to the standard DIN EN ISO 6507 using a Qness 60 Aþ hardness tester with 10 kg load (HV10). Five indents were made on each sample, from which the average Vickers hardness and standard deviation were calculated from. Furthermore, the fracture toughness (K Q ) at room temperature was determined in accordance with the standard ASTM E399 for selected heat treatment conditions.
Following the microstructural characterization and mechanical testing of the heat-treated samples, various in situ techniques were applied for investigating the microstructural changes during intercritical annealing and subsequent aging. The dilatation curve of a solution-annealed sample was recorded for annealing at 700°C for 2 h by employing the dilatometer used for the heat treatments. A direct approach for investigating the microstructural changes was performed by using HT-EBSD. Samples were prepared by cutting the dilatometer samples with 8 mm in diameter into thin plates with approximately 1 mm thickness, followed by mechanical grinding and polishing. HT-EBSD measurements were performed with a Zeiss Crossbeam 340 workstation equipped with an Oxford Instruments Symmetry EBSD detector and a TSL-Solutions HSAE-1000 heating stage. The acceleration voltage was adjusted to 30 kV and a step size of 100 nm was chosen. The data were accordingly converted and, subsequently, underwent a similar grain dilatation cleanup as the data from the ex situ EBSD measurements. Complementary to the HT-EBSD measurements, an Anton Paar HTK 1200 N hightemperature chamber was employed for capturing diffractograms during intercritical annealing and aging in a vacuum atmosphere. Both the XRD parameters and the determination of the phase fraction of austenite were similar to the ex situ XRD measurements.
Ultimately, an in-depth microstructural analysis was performed by means of TEM and a combination of TKD and APT. Samples with 100 μm thickness were prepared for the TEM measurements. The samples underwent electrolytic polishing using a Struers TenuPol-5 machine at approximately À20°C. A Struers A2 electrolyte was employed and the voltage was set to 30 V. The microstructure of the samples was then analyzed at 200 kV using a ThermoFisher Scientific Talos F200X G2 microscope equipped with a Super-X EDS system. For an accurate chemical analysis of the microstructure, APT tips were prepared by focused ion beam (FIB) milling. Prior to the transfer into the CAMECA IMAGO LEAP 3000X HR atom probe, TKD measurements at 30 kV and a step size of 5 nm were performed for obtaining crystallographic information. The details concerning the experimental setup are described in the study by Babinsky et al. [29]. The collected Kikuchi patterns were reindexed by using the NPAR function implemented in OIM Analysis v8.6 from EDAX for increasing the quality of the results. Furthermore, a grain dilatation cleanup with a minimum CI of 0.6 was applied to the EBSD data to reduce noise. Ultimately, APT measurements were conducted at a base temperature of 20 K in laser mode. The laser energy was adjusted to 0.2 nJ and a repetition rate of 200 kHz was chosen. The APT data were reconstructed and analyzed with the software IVAS 3.8 from CAMECA. The radius of the reconstructed tip for a given applied voltage V 0 can be generally expressed as where k f is the geometric field factor, which takes the shape of the tip into account, and F e is the evaporation field. [30] A standard value of 33 V nm À1 was assigned to F e , and k f was set to 2.85 for the reconstruction.

Microstructural Characterization and Mechanical Testing
In Figure 2, the EBSD results of the samples subjected to SI 700 and SI 700 A are shown. Figure 2a,b represents the inverse pole figure (IPF) color maps for martensite including grain boundaries with a misorientation angle higher than 10.5°for visualizing the martensitic block boundaries and the interfaces between martensite and austenite. Figure 2c,d shows phase color maps for austenite (red) of these microstructures, overlaid with the fit parameters to visualize the martensitic structure. The EBSD results illustrate that a large amount of reverted austenite was formed during aging after intercritical annealing at 700°C for 2 h, while almost no retained austenite was present after intercritical annealing itself. The reverted austenite grains are uniformly distributed within the martensitic microstructure, and tend to have an acicular shape along the martensitic laths. The phase fraction of austenite determined by XRD is listed in Table 3 for each conducted heat treatment. The results show that the addition of intercritical annealing (SI 700 ) results in slightly more retained austenite compared to solution annealing (S). As already revealed by the EBSD measurements, the formation of reverted austenite during aging is significantly promoted by intercritical annealing. Consequently, the total amount of austenite after aging consists mainly of reverted austenite and a small quantity of retained austenite. In addition, the results demonstrate that the temperature during intercritical annealing influences the amount of reverted austenite during subsequent aging, as lower temperatures lead to higher phase fractions. A high amount of reverted austenite (more than 10 vol%) is also present after the SI 760 SI 760 A heat treatment.
For assessing the mechanical properties, Vickers hardness testing was performed, and, for selected heat treatments, the fracture toughness at room temperature was determined. The results are also listed in Table 3 showing a decrease in hardness with the introduction of intercritical annealing, particularly for the SI 760 SI 760 A condition. The highest hardness (445 HV10) was measured on the sample subjected to the standard heat treatment (SA), which also exhibited the lowest phase fraction of reverted austenite among all aged samples. An insignificant influence on the hardness was found for the variation of the temperature during intercritical annealing. Furthermore, the results show that the fracture toughness was significantly improved from 96 to 149 MPa(m) 1/2 by introducing intercritical annealing at 760°C before aging, while the hardness exhibited only a  www.advancedsciencenews.com www.aem-journal.com moderate reduction of less than 5%. The highest fracture toughness (169 MPa(m) 1/2 ) was measured for the specimen subjected to the SI 760 SI 760 A heat treatment. However, the validation criteria for the qualification of K Q as K IC according to the standard ASTM E399 were not totally fulfilled for the latter specimen.

In Situ Experiments
Several in situ techniques were applied to study the microstructural changes during intercritical annealing and subsequent aging. Figure 3 shows the dilatation curve for intercritical annealing at 700°C for 2 h after solution annealing. As marked by the green dashed oval, the dilatation curve deviates from the linear trend during heating at approximately 550°C. As reported in the study by Kapoor et al. [31], this inflection corresponds to the precipitation of β-NiAl particles. Another deviation from the linear trend in the dilatation curve is visible at 900°C (see Figure 1) and is presumably associated with the occurrence of a two-stage austenitization during continuous heating. [31,32] Thus, martensitic regions remain in the microstructure above A f until the second step of austenitization commences at around 900°C. Moreover, the sample contraction during isothermal holding at 700°C indicates a proceeding time-dependent phase transformation of martensite into austenite. As marked by the red circle, the martensitic transformation took place at approximately 150°C upon cooling. However, the dilatation curve implies that cooling to room temperature does not result in a complete retransformation of austenite into martensite. The microstructural changes during intercritical annealing were investigated in more detail by HT-EBSD measurements. Figure 4a-d depicts the IPF color maps for martensite and the corresponding phase color maps at various stages during the SI 700 A heat treatment. Grain boundaries with a misorientation angle higher than 10.5°are highlighted in all maps. The first EBSD scan was conducted at room temperature before intercritical annealing (Figure 4a), and indicates a fully martensitic microstructure. As demonstrated in Figure 4b, a large amount of martensite transformed into austenite during intercritical annealing at 700°C after a dwell time of approximately 1 h. After cooling to room temperature (Figure 4c), the martensitic microstructure appears to be almost identical as before intercritical annealing was applied. During subsequent aging at 552°C (Figure 4d) a large amount of reverted austenite was formed after a dwell time of around 2 h, although the applied annealing temperature was fairly below A s . Moreover, it is evident that reverted austenite tends to inherit the orientation of austenite formed during intercritical annealing (compare Figure 4b,d).
For the determination of the phase fractions present during intercritical annealing and aging, HT-XRD measurements were performed. Figure 5a displays the evolution of the phase fraction of austenite for annealing temperatures of 700, 730, and 760°C, respectively. A rapid increase in the phase fraction of austenite occurred within the first minutes of annealing independent of the temperature. Furthermore, the curve corresponding to annealing at 700°C underpins that a considerable amount of martensite transformed into austenite, even though, the nominal A s is just slightly below the annealing temperature. The evolution of the phase fraction of reverted austenite during aging at 552°C for 4 h is depicted in Figure 5b. More than 35 vol% reverted austenite was formed as a result of intercritical annealing at 700°C for a dwell time of 2 h. In contrast, the phase fraction of reverted austenite reached approximately 11 vol% when solely solution annealing was performed prior to aging. In both cases, most of the austenite reversion takes place during the first few minutes, as the austenite content increases insignificantly afterward.

Transmission Electron Microscopy and Atom Probe Tomography Measurements
In-depth investigations of the microstructure after intercritical annealing at 700°C were performed using TEM. Figure 6a,b shows martensitic laths in bright-field (BF) and high-angle annular dark-field (HAADF) mode, respectively.
The corresponding EDS maps for Fe, Cr, Ni, and Mo are displayed in Figure 7. It is worth noting that elongated Ni-enriched zones of considerable size are present in the microstructure. At the same time, the exact same regions are depleted in Cr and Mo. Moreover, the HAADF image (Figure 6b) displays several bright spots, which exhibit high concentrations of Cr and Mo, but low amounts of Fe and Ni in the corresponding EDS maps in Figure 7. These particles are presumably Cr-and Mo-rich carbides, which are known to be present in PH 13-8 Mo maraging steels. [33,34] For quantitative chemical analysis of the Ni-enriched zones in martensite after intercritical annealing, APT measurements were conducted. To ensure that no retained austenite and solely martensite is analyzed, TKD was performed beforehand. Figure 8a,b shows the IPF color map for martensite and the phase color map of the analyzed tip overlaid with the image quality (IQ) values, respectively. The phase color map confirms that the investigated tip fully consists of martensite (blue), and the IPF color map reveals a martensitic grain boundary. The 3D reconstruction of the analyzed tip is depicted in Figure 8c, showing the    As shown in Figure 9, a cylindrical region of interest (ROI) with 50 nm in length was placed inside the reconstructed tip to measure to 1D concentration profile across the grain boundary in martensite. The Ni concentration is higher below the boundary, while the Cr and Mo content is lower compared to the region above. However, the average concentration of Ni and Cr approximately matches the values from the nominal chemical composition (compare with Table 1). These results quantitatively underpin that the alloying elements are unevenly distributed within the microstructure after intercritical annealing. It is noteworthy that clusters of Ni and Al ions, which  www.advancedsciencenews.com www.aem-journal.com represent intermetallic β-NiAl particles in APT 3D reconstructions of PH 13-8 Mo maraging steels, [35] were not found in the investigated tip.

Austenite Reversion during Intercritical Annealing
Depending on the heating rate and the chemical composition, the austenite reversion in Fe-Ni alloys can underlie various mechanisms. [27,[36][37][38] For example, at higher heating rates the martensite to austenite transformation tends to change from a diffusional reversion to a shear mechanism due to insufficient time for diffusion. [27] However, the proceeding martensite to austenite transformation during isothermal holding revealed by the HT-XRD measurements (see Figure 5a) as well as the measured sample contraction in the dilatometry experiments (see Figure 3) strongly indicate a diffusion-controlled character of the austenite reversion in the conducted experiments. It was found for martensitic steels with similar chemical compositions that the austenite reversion is accompanied by Ni partitioning. [39,40] To confirm that considerable partitioning of substitutional elements takes place during intercritical annealing of PH 13-8 Mo maraging steels, DICTRA simulations were carried out with the software Thermo-Calc 2022a for a simplified alloy system (Fe-12.75Cr-8Ni-2.25Mo-1.1Al) using the thermodynamic database TCFE2 and the kinetic database MOBFE5. The principles of the basic model of DICTRA are described in ref. [41]. Isothermal holding at 700°C for 2 h was simulated for a cell with planar geometry and 500 nm in length consisting of 98% martensite and 2% austenite (Figure 10a). The low amount of austenite represents retained austenite after solution  annealing. The evolution of the phase composition is shown in Figure 10b, and Figure 10c illustrates the diffusion of the main substitutional elements. The calculated concentration profiles for Cr, Ni, and Mo after 0, 10, 30, and 120 min are shown in Figure 10d-f. The simulation predicts approximately 10 vol% less austenite during intercritical annealing at 700°C compared to the experimental results from HT-XRD (Figure 5a). It is conceivable that this difference is at the root of simplifications of the model, such as the chemical composition of the system. According to the concentration profiles in Figure 10d-f, partitioning of Cr, Ni, and Mo takes place during intercritical annealing. After a dwell time of 2 h at 700°C the Ni content in austenite is somewhat above the nominal concentration, while the residual martensite is heavily depleted in Ni. The exact opposite is the case for Cr and, to a lower extent, also for Mo. The intercritical annealing temperature influences the degree of the Ni enrichment since higher or lower phase fractions of austenite require less or more stabilizing elements, respectively. Therefore, lower Ni concentration in austenite can be expected in austenite during intercritical annealing at 730 and 760°C. However, the diffusivities of Cr, Ni, and Mo atoms are even at 700°C sufficient for thorough partitioning of these substitutional elements between austenite and martensite. The low C content in PH 13-8 Mo possibly facilitates the redistribution of the substitutional elements since less interference with C atoms and, hence, higher diffusivities can be expected. [42]

Microstructure after Intercritical Annealing
As revealed by the results from HT-EBSD and dilatometry, the majority of austenite retransforms into martensite after intercritical annealing. However, compared to solution annealing (see Table 3), a higher phase fraction of retained austenite is present after intercritical annealing due to the partitioning of Ni and, therefore, a chemical stabilization of austenite. The consequences of the redistribution of substitutional atoms during intercritical annealing are also noticeable in martensite at room temperature, as TEM/EDS and APT revealed an inhomogeneous distribution of Cr, Ni, and Mo atoms. It is suggested that the zones in martensite, which are depleted in Ni, but enriched in Cr and Mo (see Figure 7), were also martensitic during intercritical annealing. On the contrary, the large areas enriched in Ni were presumably austenitic at 700°C, but transformed into martensite upon cooling.
The almost identical appearance of the martensitic structure before and after intercritical annealing hints the occurrence of the so-called austenite memory effect, i.e., reversing austenite inherits the morphology and orientation of the prior austenite. [43][44][45][46][47][48] Furthermore, the austenite memory effect can be directly observed in the HT-EBSD results where the orientation of reverted austenite (Figure 4d) appears to be identical compared to austenite formed during intercritical annealing (Figure 4b). Nevertheless, it should be taken into account that the austenite memory effect was observed on a free surface where the mechanical constraints are different compared to bulk regions. Various mechanisms are proposed for the austenite memory effect in literature. However, a diffusion-controlled austenite formation accompanied with an inheritance of geometrically necessary dislocations was recently assessed to be the most probable variant for an alloy akin to PH 13-8 Mo maraging steels at a similar heating rate (3.3 K s À1 ). [43] Although effective refinement of the martensitic structure and an increase in toughness after intercritical annealing were reported for other martensitic steels, [19,49] it is proposed that a refinement of the PAG size and, subsequently, the martensitic structure through single intercritical annealing is impeded by the austenite memory effect in PH 13-8 Mo maraging steels. Insignificant change in the PAGs was also reported in the study by Guo et al. [10] for the same steel after intercritical annealing at 760°C for 2 h. On the contrary, effective grain refinement was observed by the same authors after a double sequence of solution and intercritical annealing. However, the assessment of the evolution of grain size during the SI 760 SI 760 A heat treatment was not part of our investigations. As demonstrated in the results, no intermetallic β-NiAl precipitates were found in the 3D reconstruction of the evaporated APT tip, although, the deviation at approximately 550°C from the linear trend of the dilatation curve during heating ( Figure 3) is a strong indicator for the precipitation of intermetallic particles. [31] Therefore, it can be concluded that after annealing at 700°C throughout dissolution of these precipitates had taken place.

Formation of Reverted Austenite during Aging
The promoted formation of reverted austenite during aging at 552°C can be largely attributed to the Ni-enriched zones in martensite stemming from intercritical annealing. After solely solution annealing the substitutional elements are expected to be more homogenously distributed before aging. Consequently, Ni atoms need to be provided by dissolved intermetallic β-NiAl particles for the growth of reverted austenite. Therefore, considerable phase fractions can only be obtained at high aging temperatures or after prolonged aging times. [33] To underpin the effect of partitioning during intercritical annealing on the formation of reverted austenite during subsequent aging, the IPF color maps from the HT-EBSD results in Figure 4 are depicted selectively for austenite and martensite in Figure 11. As marked by the dashed white circles, zones inside the microstructure, which had remained martensitic during intercritical annealing, neither tended to transform into austenite during subsequent aging since chemical stabilization of martensite by partitioning of Cr, Ni, and Mo had taken place during intercritical annealing (compare Figure 11b,d). On the contrary, zones, which had already transformed into austenite during intercritical annealing, exhibited facilitated austenite reversion with simultaneous orientation inheritance during subsequent aging (compare Figure 11a,c). Nevertheless, comparing the HT-XRD results during aging (Figure 5b) with the final austenite contents (see Table 3) reveals that the majority of reverted Figure 11. IPF color maps for austenite and martensite during a,b) intercritical annealing at 700°C and c,d) subsequent aging at 552°C. The color maps underpin that the martensitic zones during intercritical annealing tend to remain martensitic during following aging.
www.advancedsciencenews.com www.aem-journal.com austenite formed at 552°C is not thermally stable after cooling to room temperature. This also applies to the conventional heat treatment without additional intercritical annealing (SA). Furthermore, Figure 11 implies that two types of tempered martensite are present in the microstructure after aging. On the one hand, martensite, that neither transformed into austenite at 700°C nor during aging, was tempered twice. On the other hand, there is austenite that had transformed into martensite after intercritical annealing, but did not reverse into austenite at 552°C afterward. In addition, a considerable amount of fresh martensite is present after aging since a large quantity of reverted austenite is not thermally stable at room temperature. Ultimately, the final microstructure mainly consists of three different types of martensite, reverted austenite and intermetallic β-NiAl precipitates after the three-step heat treatment. It was reported that the combination of fresh and tempered martensite is the root cause of the toughness improvement since such a dual-phase martensite is expected to act similarly compared to a combination of martensite and retained/reverted austenite. [10] However, the results of the present work strongly imply that the enhanced fracture toughness due to additional intercritical annealing can be mainly attributed to the increased amount of reverted austenite. It is well-proven that this phase effectively contributes to the toughness and ductility of PH 13-8 Mo maraging steels. [9,50] The lower the intercritical annealing temperature, the more reverted austenite is formed during subsequent aging, thus, improved fracture toughness can be expected. Nevertheless, no significant variation in Vickers hardness was registered after conducting intercritical annealing at different temperatures. This underpins that intermetallic β-NiAl precipitates provide the dominant strength contribution to PH 13-8 Mo maraging steels since the same aging conditions were applied to all samples. For the reason that intercritical annealing enables to obtain high phase fractions of reverted austenite at lower aging temperature, finer intermetallic β-NiAl precipitates and less softening of the martensitic matrix can be expected in comparison to overaging, i.e., aging at higher temperatures or for prolonged times. Ultimately, an overall improvement of the strength/ductility combination is proven by this three-step heat treatment strategy. SI 760 SI 760 A is an exception since the hardness was drastically lowered by this heat treatment. In comparison to SI 760 A, SI 730 A, and SI 700 A, no direct correlation between the hardness and the phase fraction of reverted austenite was registered. It is conceivable that the low hardness after this heat treatment is at the root of the martensitic matrix or the β-NiAl precipitates. However, a detailed microstructural characterization of the respective samples was beyond the scope of this study.

Conclusion
The present study provides insight into how the implementation of intercritical annealing after solution annealing and prior to aging affects the microstructure of PH 13-8 Mo maraging steels. Several in situ techniques were applied to directly investigate microstructural changes during various stages of the three-step heat treatment, and were complemented by high-resolution characterization methods of the microstructure after intercritical annealing. In addition, the influence on the hardness and fracture toughness was studied. The following conclusions can be drawn: 1) Partitioning of the substitutional elements Cr, Ni, and Mo takes place during intercritical annealing, whereby austenite becomes enriched in Ni, while Cr and Mo atoms diffuse into martensite. The stabilization of austenite provokes a higher phase fraction of retained austenite at room temperature compared to the solely solution-annealed state; 2) Due to partitioning of Cr, Ni, and Mo atoms during intercritical annealing, zones, that are enriched in Ni, but depleted in Cr and Mo, remain in the microstructure after the martensitic transformation. These Ni enrichments promote the formation of reverted austenite during subsequent aging, as the Ni redistribution is decisive for the growth of this phase. The lower the intercritical annealing temperature, the more pronounced the Ni enrichments in martensite are. Ultimately, lower temperatures during intercritical annealing result in higher phase fractions of reverted austenite after aging; 3) Improved toughness was registered when intercritical annealing had been introduced. This can be mainly attributed to the higher phase fraction of reverted austenite in comparison to the standard heat treatment consisting of solution annealing and aging (SA). A toughness contribution stemming from grain refinement is ruled out since the austenite memory effect occurs during intercritical annealing and leads to an akin martensitic structure before and after this annealing step. However, an additional beneficial contribution to the fracture toughness possibly results from the combination of fresh and tempered martensite; and 4) Intercritical annealing enables the formation of high-phase fractions of reverted austenite at moderate aging temperatures. Compared to overaging, finer intermetallic β-NiAl precipitates and less softening of the martensitic matrix are expected. Consequently, an improved combination of strength and toughness is possible by using this three-step heat treatment.