Hybrid Tool Holder by Laser Powder Bed Fusion of Dissimilar Steels: Towards Eliminating Post-Processing Heat Treatment

Abstract

The hybridization of additive manufacturing (AM) with conventional manufacturing processes in tooling applications allows the customization of the tool. Examples include weight reduction, improving the vibration-dampening properties, or directing the coolant to the critical zones through intricate conformal cooling channels aimed at extending the tool life. In this regard, metallurgical challenges like the need for a post-processing heat treatment in the AM segment to meet the thermal and mechanical properties requirements persist. Heat treatment can destroy the dimensional accuracy of the pre-manufactured heat-treated wrought segment, on which the AM part is built. In the case of dissimilar joints, heat treatment may further impact the interface properties through the ease of diffusional reactions at elevated temperatures or buildup of residual stresses at the interface due to coefficient of thermal expansion (CTE) mismatch. In this communication, we report on the laser powder bed fusion (L-PBF) processing of MAR 60, a weldable carbon-free maraging powder, to manufacture a hybrid tool holder for general turning applications, comprising a wrought segment in 25CrMo4 low-alloy carbon-bearing tool steel. After L-PBF process optimization and manipulation, as-built (AB) MAR 60 steel was characterized with a hardness and tensile strength of ~450 HV (44–45 HRC) and >1400 MPa, respectively, matching those of pre-manufactured wrought 25CrMo4 (i.e., 42–45 HRC and 1400 MPa). The interface was defect-free with strong metallurgical bonding, showing slight microstructural and hardness variations, with a thickness of less than 400 µm. The matching strength and high Charpy V-notch impact energy (i.e., >40 J) of AB MAR 60 eliminate the necessity of any post-manufacturing heat treatment in the hybrid tool.

1. Introduction

Additive manufacturing (AM) is a promising manufacturing technique in the tooling industry. AM allows the manufacture of complex tool bodies with intricate conformal cooling channels that cannot be achieved via conventional manufacturing processes [1,2]. Laser powder bed fusion (L-PBF) is an AM technology suitable for the manufacture of tools and components [3,4,5,6]. L-PBF-processed components show high precision and reproducibility, and part-to-part variations on one build job are reported to be of a Gaussian nature, excluding systematic or operator contribution [7,8]. Further, it offers negligible variations in view of batch-to-batch repeatability. Moreover, as a result of limited layer thickness (i.e., several tens of micrometers), the dimensional accuracy of features and wall thickness often show a negligible variation from the specified values [7,8,9].

The hybridization of additive manufacturing with conventional manufacturing processes in tooling applications allows the production of geometries that require fewer manufacturing steps, resulting in shorter lead times and faster deliveries [3,4,5,6,10]. Further, it adds flexibility to the application of dynamic changes and innovative designs compared to conventional processes (i.e., customization) [11]. Examples of such customizations include weight reduction through topology optimization, improving the vibration-dampening properties, or directing the coolant to the cutting edge through intricate conformal cooling channels to extend the tool life [11]. Hybrid tool bodies are composed of an additively manufactured segment, where the use of complex designs, intricate conformal cooling channels, and compositionally varied materials in the same geometry, improve the performance. This segment is printed on a pre-manufactured conventional tool holder back-end, sometimes referred to as a blank (hereafter referred to as back-end). Examples are hybrid extrusion dies comprising mandrel tips with complex cooling channels printed on a wrought tool bridge, hybrid hot stamping tools, and tool holders for general turning [12,13].

Tool holder bodies for general turning or milling are normally manufactured using low-to-medium-carbon martensitic tool steels as they offer high hardness, wear, and fatigue resistance in combination with adequate toughness [14,15]. Back-ends of C-bearing martensitic tool steels cannot, however, be reliably combined with laser AM processes. First, L-PBF of medium-carbon tool steels is still challenging due to the risk of cold cracking, in agreement with their poor weldability [16]. Second, a pre-manufactured conventional tool holder back-end is normally in a quenched and tempered condition. However, an L-PBF-processed part, if printed in the C-bearing tool steel, is generally extremely hard, with low toughness, which are characteristics of quenched martensite [17,18], and must undergo a final heat treatment to achieve the required performance by balancing these properties [18]. Such heat treatment, involving prior quenching or even direct tempering, would destroy the dimensional stability of the pre-manufactured back-end on which the additively manufactured part is built. This is due to the dimensional changes and distortions affiliated with both the quenching and tempering of carbon martensite [19,20,21]. Maraging steels, on the other hand, are a weldable class of ultrahigh-strength C-free martensitic steels, which can be welded without preheating either in the annealed or heat-treated condition [22,23]. Because of this, they have become the most popular ultrahigh-strength steel of choice in laser AM [1,24,25]. Moreover, they show negligible dimensional changes upon hardening and aging [22,23]. The martensitic microstructure of maraging steels is normally achieved through a high nickel content (i.e., over 5 wt.%) [23,26]. Fe-Ni martensite in the quenched or as-built (AB) state is much softer (i.e., ~350 HV10) than medium-carbon tool steels, achieving an extremely hard Fe-C martensitic microstructure through interstitial carbon in the supersaturated martensite (i.e., ~500–700 HV10) [1]. Therefore, parts additively manufactured from carbon-free steels, such as maraging steels (e.g., 18Ni300), must be subsequently subjected to aging to achieve high hardness levels (400–700 HV10) through intermetallic precipitation [13]. This aging treatment, if realized at a higher temperature than that of the tempering of the pre-manufactured back-end, damages the microstructure stability and strength of the pre-manufactured back-end on which the additively manufactured part is built. Moreover, normally, heat treatment of dissimilar joints may introduce the precipitation of unwanted phases at the interface due to the elemental concentration gradients and enhanced inter-diffusion kinetics at elevated temperatures.

In this communication, we report on the use of highly alloyed Osprey^® MAR 60 steel powder, which has similar chemistry as the conventional 13Ni400 maraging steel [25,27,28,29]. Like other maraging systems, MAR60 shows excellent weldability and L-PBF processability, with an achievable hardness of around ~450 HV10 in the AB condition without any aging [28,30]. Due to its relatively high hardness, which is in the range of heat-treated or induction-hardened low-alloy carbon-bearing tool steels, this steel can be used for the manufacture of a variety of hybrid tools and tool holders in general turning, with no need for additional heat treatments. Strengthening in the AB state (i.e., not aged) is achieved thanks to an extremely high solid solution strengthening by the substitutional element Mo (~10 wt.% in chemistry) and dislocation density through the out-of-equilibrium rapid solidification and cooling, characteristic of the L-PBF process [29,30]. Ni (13 wt.%) and Co (15 wt.%) are adjusted to achieve a martensitic microstructure by keeping the nominal martensite start (Ms) temperature higher than ~200 °C [30,31]. Further, we report on the variations in AB hardness (ranging from 370 HV1 to 460 HV1) affected by the duration of interlayer deposition time. This variation is particularly alleviated by manipulating the interlayer deposition time to achieve a uniform and high AB hardness (~450 HV), matching that of the back-end wrought segment. Finally, the interface characteristics of the dissimilar joint are studied by means of microstructural and hardness analysis.

2. Materials and Methods

2.1. Powder and Back-End Tool Holder Material

According to the manufacturer’s (Sandvik Osprey, Neath, UK) datasheet and powder certificate, gas-atomized Osprey^® MAR 60 maraging powder (composition shown in

Table 1. Chemical composition of the MAR 60 powder, in wt.%.

	Co	Ni	Mo	Ti	Cr	Mn	Si	Al	C	Fe	O	N
Nominal	15.00	13.00	10.00	<0.30	<0.20	<0.1	<0.1	<0.1	<0.03	Bal.	<0.05	<0.04
Powder	14.80	13.10	9.90	0.20	0.13	0.05	0.05	0.06	0.01	Bal.	0.04	0.01

) with a particle size distribution (PSD) of 15–45 µm was used to additively manufacture the samples (tool bodies) on a back-end of a Seco-Capto^TM tool holder with a flange diameter of 50 mm. The spherical powder morphology, which is characteristic of the gas atomization process, is shown in scanning electron microscope (SEM) micrographs in Figure 1. A Renishaw RenAM 500 series L-PBF machine equipped with a 500 W Yb-fiber laser was employed for this purpose.

The back-end was in a commercially provided 25CrMo4 low-alloy tool steel with a chemical composition of 0.25% C, 0.2% Si, 0.7% Mn, 1% Cr, 0.2% Mo, in wt.%. This steel is normally delivered in a quenched and tempered (QT) or induction-hardened condition [32]. In the current work, 25CrMo4 steel was delivered in a QT condition, with a hardness of ~430 HV10 and a tensile strength of ~1400 MPa.

2.2. Additive Manufacturing

Process optimization was performed by printing cubes (15 × 15 × 15 mm³) by employing a stripes scanning strategy with 67 degrees of rotation between the layers, with the layer thickness (t), and hatching spacing (h) being set to 60 µm and 90 µm, respectively. For the design of experiments (DoE), a factorial design with two factors (i.e., laser power (P) and scanning speed (v)) was considered. The laser power was varied from 250 W to 350 W, with five levels (i.e., 250, 275, 300, 325, and 350 W), while the scanning speed was varied from 800 mm/s to 1200 mm/s, comprising five levels (i.e., 800, 900, 1000, 1100, and 1200 mm/s). In this regard, a laser power of 300 W, a hatching spacing of 90 µm, and a laser scanning speed of 800 mm/s were chosen as the parameter set, resulting in an almost fully dense microstructure. The above parameters resulted in a volumetric energy density (VED) of 69.5 J/mm³ and an area energy density (E_h) of 4.2 J/mm² according to Equations (1) and (2), respectively [33].

$$VED \left[{\displaystyle \frac{\mathrm{J}}{{\mathrm{m}\mathrm{m}}^3}}\right]={\displaystyle \frac{P}{t.h.v}}$$

(1)

$$E_h \left[{\displaystyle \frac{\mathrm{J}}{{\mathrm{m}\mathrm{m}}^2}}\right]={\displaystyle \frac{P}{h.v}}$$

(2)

2.3. Microstructure and Mechanical Characterization

Porosity measurements were carried out using image analysis on the stitched light optical microscope (LOM) images at 10× magnification, captured on carefully ground and polished (to 3 µm cloth) metallographic cross-sections (15 × 15 mm²). Microstructural analysis was performed using LOM and a high-resolution field emission gun scanning electron microscope (FEG-SEM) on metallographic cross-sections. Energy-dispersive X-ray spectroscopy (EDS) spot, line scan, and elemental mapping, as a semi-quantitative analysis, were used for the determination of variations in the chemical compositions. Electron Backscatter Diffraction (EBSD) measurements were carried out using Nordlys F detector equipment in FEG-SEM. X-ray diffraction (XRD) was carried out using a Bruker D8 Advanced instrument equipped with a Co Kα (λ = 0.179 nm) source (7-peak emission profile). The data were analyzed using Topas 5.2 software from Bruker AXS.

Vickers hardness measurements, using loads of 5 kg.f (HV5) and 10 kg.f (HV10), were carried out following ASTM E92-17 [34]. A minimum of five indentations were made, and the average value was reported for the global hardness of the bulks. Microhardness profiles were measured using 1 kg.f load (HV1). Tensile tests were carried out at room temperature using both vertically and horizontally fabricated cylindrical specimens, measuring 45 mm in height and 6 mm in diameter, which were machined to comply with the ASTM E8/E8M-22 standard [35]; three samples were tested for each build orientation. Additionally, instrumented Charpy V-notch (CVN) tests were performed following ASTM E23-18 guidelines [36]. For this purpose, samples were initially printed with dimensions of 55 × 12 × 12 mm³ and then machined to meet standard requirements. Impact properties were tested for both vertically and horizontally built specimens; three samples were tested for each build orientation.

2.4. Thermodynamic Simulations

Thermodynamic calculations were made using Thermo-calc software (TCFE12, MOBFE7 database). Schiel solidification modeling was conducted by considering solute trapping based on the Aziz model [37], with solute trapping in primary solid phase only, which is suggested to be suitable for high solidification rates [38]. Further, the model calculates the solid/liquid interface velocity (solidification rate, V_s) using the product of the scanning speed (V_scanning, here set to 800 mm/s) and the cosine of the angle (α) between solid–liquid boundary and the scanning direction (set to 1 in this case). In Fe-Ni-Co-Mo quaternary alloys, the solidification starts with austenite, and this phase is the only stable phase down to martensite start temperature of the alloy upon rapid cooling [23]. Therefore, austenite was set as the primary phase. To elaborate on the elemental micro-segregation, DICTRA simulations were coupled with Scheil calculations. A simulation domain size of 500 nm (i.e., the average radius of the cellular solidification structure in this work) was employed to simulate the micro-segregation profiles from the core of the cellular structure to the intercellular micro-segregation (i.e., cell boundary). Finally, a property module was used to calculate the variations in martensite start (Ms) and finish (Mf) temperatures as a function of elemental micro-segregation.

3. Results and Discussion

3.1. Microstructure of MAR 60 in As-Built Condition

An example of the final processing window, showing the porosity (%) and hardness (HV5) vs. VED and E_h, is shown in Figure 2a, and examples of LOM micrographs of the bulks processed with non-optimum and optimized parameters are shown in Figure 2b and c, respectively. Hardness increased by reducing the porosity, and the maximum hardness in samples with a relative density of >99.95% was around 440–450 HV5 (Figure 2a).

Microstructural features and phase constitutions of the as-built specimen in MAR60 are shown in Figure 3.

In general, the L-PBF-processed samples showed traces of melt pool boundaries (etched by Nital 2% reagent) and a martensitic microstructure with columnar prior austenite grains along the build direction (heat dissipation direction) when etched by Kalling’s reagent (Figure 3a). A fine cellular/dendritic microstructure and columnar PAGs passing through the melt pools are shown in an SEM micrograph in Figure 3b. The cellular boundaries (brighter in a back-scatter-mode SEM micrograph in Figure 3c) were depleted of Fe and Co and were rich in the element Mo, with very slight variations in Ni, given the EDS line profile analysis in Figure 3d. The strong micro-segregation of alloying elements is due to constitutional undercooling during the out-of-equilibrium rapid solidification in L-PBF [39]. XRD results (Figure 3e) confirmed the existence of retained austenite (RA) to ~10 vol.% in the microstructure. RA was only found at the intercellular micro-segregation (i.e., cellular boundaries), as shown by EBSD maps (Figure 3f–h). In agreement with the literature [25,40], the crystal orientation of RA is dependent on the orientation of the parent prior austenite grain (highlighted in the inverse pole figure map in Figure 3h by arrows and a dashed line). This confirms the assumption that a drop in the martensite start (Ms) temperature at the cellular boundaries, as a result of heavy micro-segregation, does not allow the completion of martensitic transformation upon cooling within those regions in parent austenite, hence stabilizing RA.

To understand the micro-segregation behavior in this alloy, thermodynamic simulations were conducted. Figure 4a shows the Scheil solidification modeling, considering solute trapping. Solidification starts with the formation of austenite (FCC_A1) as the primary phase. The out-of-equilibrium solidification path deviates at around 0.5 mole fraction of solid from that of equilibrium, indicative of micro-segregation, which is accompanied by a drop in the melting point of an enriched liquid, thus widening the solidification temperature range. By coupling the Scheil simulation and diffusion module (DICTRA), it is possible to predict the compositional gradients from the center of the cell (the first solid to form) to the cellular boundaries (i.e., the micro-segregated region or the last drop of liquid to solidify). For this purpose, a solidification cell size of 1 µm (half-cell width of 0.5 µm) was chosen according to the SEM micrograph in Figure 4b, where no chemical etching was conducted, to correctly reveal the actual size of the cell as well as the true thickness of the maximum intercellular micro-segregation. The concentration profiles are shown in Figure 4c. In agreement with the experiments, a simulation confirmed the heavy micro-segregation of Mo as well as Ti, although Ti wt.% in MAR 60 was limited to 0.2 wt.% and its micro-segregation could not be highlighted in EDS line scans. Moreover, Ni only showed a slight increase, while Co was slightly depleted at the cell boundaries, in agreement with the EDS line analysis. Consequently, the concentration profiles along the cell core to the cell boundary were used as input in the property module in Thermo-calc to plot the martensite temperatures (Ms and M_f) as a function of the distance from the cell center (Figure 4d). In agreement with the EBSD results, it is shown that Ms and M_f temperatures drop heavily at the micro-segregated regions. Specifically, at around 440 nm from the cell center, M_f drops to below 20 °C, indicating that martensitic transformation cannot be completed and that some RA can be stabilized. At maximum micro-segregation (i.e., cell boundary), even Ms is as low as ~40 °C. In this calculation, the relative thickness of the heavily micro-segregated zone (~60 nm) to that of the half-cell width (500 nm) is around 12%. This number is close to the RA vol.% measured by XRD (i.e., 10%).

3.2. Mechanical Properties of MAR 60 in As-Built Condition

Tensile curves are reported in Figure 5a, and the respective properties are shown in

Table 2. Mechanical properties of MAR60, 25CrMo4 steel, and wrought 13Ni400 steel.

Sample/Orientation	Yield Strength (MPa)	Tensile Strength (MPa)	E-Modulus (GPa)	Elongation (%)	Area Reduction (%)	Hardness [HV10]	CVN Impact Energy [J]	Ref.
MAR 60 vertical	1301 ± 13	1405 ± 6	190 ± 6	16.9 ± 0.2	63 ± 3	442 ± 6	45 ± 1	This work
MAR 60 horizontal	1279 ± 22	1423 ± 5	204 ± 26	16.8 ± 0.2	58 ± 1	450 ± 10	44 ± 2	This work
Back-end 25CrMo4	1150	>1350	-	>6	-	430 *	20	Supplier
13Ni400 (hot-rolled)	1055	1258	-	19.0	70	424 *	-	[31]
13Ni400 (annealed)	724	1145	-	19.0	72	363 *	-	[31]

* Converted from Rockwell C (HRC) hardness.

. In both vertically and horizontally built samples, yield strength was ~1300 MPa, and tensile strength was ~1400 MPa. An elongation of ~16% and an area reduction of ~60% indicate a ductile fracture. This is confirmed by looking at the fracture surface, characterized by many dimples, indicative of a trans-granular ductile fracture (Figure 5b). Hardness was in the range of ~440 HV10 at a CVN impact strength of ~45 J. The as-built L-PBF samples, benefiting from far-from-equilibrium rapid solidification, inducing a large dislocation density and high matrix supersaturation, show an even higher strength compared to a hot-rolled wrought 13Ni400 steel (i.e., UTS of 1258 MPa and hardness of 420 HV) and is much higher than that of the solution-annealed one (UTS of 1145 MPa and hardness of 360 HV) [31]. These strength levels are larger than the as-built strength of the most commercially exploited 18Ni300 grade (~1000–1250 MPa) [1,24,41,42,43].

The higher tensile strength in AB MAR60 compared to AB 18Ni300 can be a direct consequence of larger matrix supersaturation by the substitutional element Mo (10.0 wt.% in composition, ~6.2 at.%), which is two times higher than that of 18Ni300 (~4.75wt.%, 2.8 at.%). Mo and Ti are known to have a larger solid solution strengthening effect in the Fe matrix, with solid solution strengthening constants of 2362 and 2628 MPa/at, respectively, taking into consideration their high lattice and modulus misfit with respect to the iron matrix [44]. However, Co and Ni show much lower solid solution strengthening constants (i.e., 708 and 50 MPa/at, respectively) [44,45,46]. According to the solid solution strengthening equation (Equation (3)), where B_i (MPa/at) is the strengthening constant of element i and x_i,α’ is the atom fraction of element i in the matrix, differences in Mo (~6.2 vs. 2.8 at.% in 18N300) and Ti (~0.2 vs. 0.7 at.% in 18Ni300) can result in an over 100 MPa increment in the as-built strength of MAR60 compared to 18Ni300.

$${\sigma }_{ss}={\sum }_i\sqrt{B_i^2{x}_{i,{\alpha }^{\prime }}}$$

(3)

Nevertheless, from a practical viewpoint, the tensile and hardness properties of as-built MAR 60 match those of low-alloy C-bearing steels, specifically the 25CrMo4 alloy used in this work (Table 2), which normally operates at ~420–440 HV hardness levels in a QT condition. Impact energy is much higher for MAR60 at this hardness level compared to 25CrMo4 (i.e., 45 J vs. ~20 J). This facilitates the use of MAR60 in hybrid tool production by avoiding the need for any post-processing thermal treatment.

3.3. Hybrid Tool Holder

3.3.1. Cone Geometry Experiments

In engineering components and complex geometries, the cross-sectional area of the components varies as a function of the build height, and this may result in different thermal histories as a result of changes in the time elapsed when melting each layer (i.e., interlayer time). In other words, at larger cross-sections, a certain volume within the deposited layer has more time to be rapidly cooled and to dissipate heat before being reheated by the deposition of the next layer. The interlayer time becomes shorter by processing smaller cross-sectional areas, leading to larger heat accumulation. In maraging steels, where the kinetics of transformations (e.g., precipitation, austenite reversion, and annealing) are extremely rapid, the effect of cyclic heating and layer deposition times becomes more significant in view of their role in changing the microstructure and properties [47,48].

To evaluate the effect of rapid cooling and cyclic heating on the L-PBF-processed MAR 60 steel, a cone-shaped bulk was printed (Figure 6a). Cone geometry makes it possible to achieve different interlayer deposition times in a continuous way (i.e., the time taken for printing one layer, in addition to the time for the re-coater to deposit a new powder bed). This is due to the continuous decrease in the cross-sectional area to be printed at each layer in a cone geometry. This is a quite simple tool for evaluating the effect of interlayer deposition time on the heat accumulation in the solidified layers and, consequently, its possible effect on the microstructure and hardness in the as-built condition (Figure 6b).

Variations in layer printing time and the corresponding hardness as a function of build height are shown in Figure 6c and d, respectively. Starting from the bottom of the cone with interlayer times higher than ~30 s, hardness is measured at 450 HV1, similar to those achieved in the process optimization shown in the experimental part. When the interlayer time gradually decreases from ~30 s to ~20 s by increasing the height and decreasing the cross-section of the cone, hardness shows a gradual drop from 450 HV1 to around 375 HV1. By further decreasing the interlayer time (cross-section area), hardness becomes constant at 360–370 HV1. At the very top layer, hardness is recovered to 450 HV (shown by an arrow).

This behavior may be discussed in view of the heat accumulation in the part. At the bottom layers, where the interlayer time is large, the printed cross-section can be effectively cooled before the next layer is deposited. Rapid solidification and cooling may result in large dislocation densities and the achievement of highly supersaturated martensite. When the interlayer time becomes shorter as a result of the reduction in the cross-section area, the cooling time for the deposited layers becomes shorter before the next layer is deposited. The lack of sufficient cooling leads to heat accumulation, which may result in an in situ annealing of the deposited layers. Annealing may induce dislocation recovery as well as reduce matrix supersaturation in Mo upon slower cooling, leading to softening. An equilibrium phase diagram (Figure 7a) shows that due to the high Mo content in this alloy, the formation of the Fe₂Mo Laves phase is expected at temperatures as high as 900 °C; therefore, it is probable that at short interlayer times and under large heat accumulation, annealing and some Laves-phase precipitation take place. Indeed, the hardness values (i.e., 360–370 HV1) at rapid printing times, as shown in Figure 6b,c, are on par with the hardness of annealed 13Ni400 (please see Table 2) and are in agreement with authors’ findings that the solution annealing of additively manufactured MAR 60 led to softening, and the precipitation of the Laves phase upon cooling. (Figure 7b) [49].

At the very top layers, where no additional reheating is expected, the few topmost layers may be quenched without undergoing cyclic heating; consequently, hardness shows an abrupt increase at those regions and approaches ~450 HV1. To mitigate such variations, the printing time per layer should be controlled based on the required hardness. To produce the hybrid tool, the printing time per layer was artificially manipulated to be higher than 40 s. For smaller cross-sections, which resulted in shorter printing times per layer, a “dead time” was applied to the laser by making it idle before the deposition of the next layer. It is worth mentioning that in practice, the dead time can be only applied locally in printing the load-bearing “critical “areas in the tool, where maximum hardness is needed (e.g., insert seat). The bulk can therefore be processed without any interruptions in printing time, regardless of the achieved hardness. Indeed, a softer core may even result in increased toughness, which may be helpful in reducing the risk of the catastrophic failure of the tool.

3.3.2. Hybrid Tool Holder: Interface Characteristics

The hybrid tool images are presented in Figure 8a,b. The back-end in the wrought 25CrMo4 low-alloy tool steel and the AM MAR60 are addressed in Figure 8a. The cutting insert mounted on the MAR60 insert seat is also highlighted in Figure 8b.

A LOM micrograph on the interface (Figure 8c) shows stronger etching on the tempered low-alloy steel with a tempered martensite microstructure. There seems to be a microstructure change with a thickness of ~50 µm in the 25CrMo4 steel below the traces of melt pools of MAR60 at the interface (region I). Above this layer, the initial melt pools show a mixing of the two materials in a liquid state, revealed by stronger etching within the melt pools, where the fluid flow patterns represent the thermocapillary convection known as the Marangoni effect (region II, ~120 µm in thickness). Above this layer, the MAR 60 microstructure seems to be intact. An SEM micrograph coupled with EDS mapping (Figure 8d) shows quite strong bonding between the alloys, with no micro-defects. It also highlights the characteristics of region I more clearly, with a seemingly finer grain size. Moreover, looking at the C concentration (true map), a slight enrichment in this zone is evident. However, the concentration of Ni, Co, and Mo (those from MAR60) seems to be negligible.

In region II, as shown by EDS mapping and the SEM image, the mixing of the alloys in a liquid state is confirmed. At these regions, some areas within the melt pool are depleted of Ni and Co at the expense of a larger Cr concentration (from 25CrMo4 steel) and, of course, a larger Fe (matrix element in both alloys) concentration, which is not reported in the maps.

Finally, according to the EDS mapping and point analysis, the elongated non-metallic inclusions in 25CrMo4 are MnS, typically found with such morphology in hot-rolled steel [50,51,52]. These inclusions seem to disappear at the border of 25CrMo4 and region I.

The EBSD results (IPF-X map in Figure 9a) confirm a different grain morphology and size compared to the core 25CrMo4. In 25CrMo4, equiaxed PAGs highlighted by high angle boundaries are evident (>15°), and the lath martensite hierarchy (i.e., lath martensite with a parallel alignment of blocks within each packet) is clear. Please note that the step size in this EBSD mapping is large (500 nm), and the resolution of fine features, especially those of block and lath boundaries, may be affected. However, the results confirm a typical low-alloy martensitic carbon steel microstructure. Within zone I, the microstructure seems to be much finer in size, in agreement with the SEM images; moreover, in this zone, quite a large amount of RA is evident (phase map in Figure 9b). In the MAR60 region, the PAGs seem to be columnar, with a martensitic microstructure.

The EBSD results, showing a fine grain structure, coupled with the elemental mapping (Figure 8), confirm that zone I is a remelted and rapidly solidified area in 25CrMo4. RA stabilization can be due to the much finer PAGs in this zone, as well as enhanced C diffusion (fast diffuser) due to the concentration gradient between MAR60 (C-free) and that of 25CrMo (0.25% C) in the remelted zone. Please note that the RA thickness in MAR60 was in the range of 100 nm, as shown in detail in Figure 3. Therefore, the absence of RA in the MAR60 region in the current EBSD is logical because the scanning step size (500 nm) is much larger than the RA feature size in MAR60.

3.3.3. Hybrid TOOL HOLDER: Hardness

SEM images of the area near the interface, subjected to microhardness testing, and the corresponding hardness profile are shown in Figure 10. The hardness in MAR60 (Figure 10a,c) shows a quite stable value of 440–450 HV1. Within MAR60, at around 250 µm from the interface, an increase in hardness to ~460 HV1 is evident, which can be correlated to zone II (see Figure 8), which is characteristic of the mixing of 25CrMo4 and MAR60 in liquid state followed by rapid cooling (Figure 10b,c). This is followed by quite a strong increase in hardness to ~520 HV1 at the fine-grained region (zone I), confirming that this area is a rapidly quenched carbon martensite (Figure 10b,c). A sudden drop to around 410 HV1 when entering the 25CrMo4 steel, along with a limited thickness of 100 µm, suggests a very small heat-affected zone in 25CrMo4. As shown in Figure 10a,c, at a distance of ~250 µm from the interface, the hardness plateaus at ~420–430 HV1 in 25CrMo4 steel, which is the nominal hardness of a conventional back-end. The reference lines show the allowable hardness tolerance in the conventional material. Overall, the results suggest that the interface is defect-free, with strong bonding, and the microstructure and hardness variations are limited to a maximum thickness of ~400 µm thanks to the very localized melting in the L-PBF process.

4. Conclusions

In this work, a hybrid tool holder for general turning, comprising a pre-manufactured wrought segment (back-end) in low-alloy tool steel (25CrMo4), and an AM segment using MAR 60 maraging steel was manufactured. The following conclusions were made.

• As-built MAR60 maraging steel was characterized by a martensitic microstructure and ~10 vol.% intercellular retained austenite as a result of the heavy micro-segregation of Mo.
• MAR60 showed an as-built hardness of ~440–450 HV10, tensile strength of over 1400 MPa, and CVN impact energy of 40 J, properties that match those of pre-manufactured heat-treated low-alloy tool steels generally used in the manufacturing of tool holders (i.e., 420–440 HV, ~1400 MPa, and 20 J, respectively). This eliminates the need for any post-processing heat treatment of the hybrid tool.
• The high hardness and strength of AB MAR60 were the result of a large dislocation density and matrix supersaturation due to the rapid solidification in L-PBF. Additionally, a high wt.% of Mo in chemistry (i.e., 10%) further induced larger solid solution strengthening compared to other maraging systems.
• It was shown that hardness was dependent on the interlayer deposition time in L-PBF, where short interlayer times led to a softening to ~360 HV1, presumably due to the in situ annealing because of heat accumulation and insufficient cooling.
• The hybrid tool was manufactured by manipulating the interlayer deposition time, keeping it higher than 40 s for all 2D cross-sectional areas of the component. This led to a uniform hardness of ~440–450 HV1.
• The interface of dissimilar steels showed strong metallurgical bonding with no apparent defects in terms of microcracks or unwanted phases. Further, the interface showed slight microstructural and hardness variations with a limited thickness of less than 400 µm thanks to the very localized melting in L-PBF processing.
• The use of steels with tailored chemical compositions to ensure printability (weldability) and acceptable as-built mechanical properties may pave the way to eliminate the post-processing heat treatment step in the manufacture of hybrid tools by AM, especially for the components that are not subjected to elevated temperatures during the service. The use of as-built material at elevated temperatures may not be feasible due to the in situ tempering/aging, resulting in dimensional changes and mechanical properties variations.