Manufacturing oxide-dispersion-strengthened steels using the advanced directed energy deposition process of high-speed laser cladding

In this work, we demonstrate the feasibility of manufacturing an iron-based oxide-dispersion-strengthened (ODS) PM2000 composite material with the chemical composition of Fe20Cr4.5Al0.5Ti + 0.5Y2O3 (in wt.%) via the advanced directed energy deposition (DED) process of high-speed laser cladding (HSLC). The characteristic high solidification rates of HSLC processes allow the successful dispersion of nano-scaled yttrium-based oxides in the ferritic stainless steel matrix. The effective suppression of nano-particle agglomeration during the melting stage, which is frequently observed in conventional DED processes of ODS materials, is reflected by smaller dispersoid sizes and corresponding higher hardness of manufactured specimen compared to DED-manufactured counterparts.


Introduction
Oxide-dispersion-strengthened (ODS) materials are promising candidate materials for new GenIV nuclear reactors due to their tolerance toward irradiation-induced defects [1][2][3]. In addition, this material class offers superior mechanical properties at room and elevated temperatures [4][5][6][7]. Therefore, ODS materials are also considered for applications in the fields of stationary gas turbines or combustion engines in the aerospace industry [8].
ODS materials gain their outstanding mechanical properties typically from the presence of sub-micrometer-sized oxide particles that are homogeneously distributed in a corrosion-resistant metallic matrix. Manufacturing of ODS steels is typically performed via a mechanical alloying process by ball milling [9,10], followed by compaction via hot extrusion or hot isostatic pressing (HIP), subsequent hot rolling, and subtractive manufacturing [11,12]. Apart from this powder metallurgical route, only few alternative concepts enabling the successful dispersion of nanometersized oxides in a metallic matrix have been reported, such as the gas atomization reaction synthesis followed by internal oxidation reaction during HIP [13] or in-situ oxidation of a titanium wire introduced into a steel melt under severe electromagnetic stirring [14]. Other promising approaches are based on gas synthesis formation of oxides in a melt pool, internal oxidation [15,16], spark plasma sintering (SPS) [17,18]. However, most of these processes only allow the production of simply shaped bodies and need costly subtractive manufacturing before being applied in a particular application.
Laser additive manufacturing (LAM) is principally capable of producing highly complex parts directly from computer files and raw material powders. LAM processes are characterized by small melt pool dimensions with high solidification rates in combination with complex flow patterns due to the so-called marangoni forces or with the formation 1 3 of a keyhole due to the displacement flow, resulting in effective dispersion of nano-scaled oxide particles. In the literature, the feasibility of the LAM-method of laser powder bed fusion (L-PBF), also termed as selective laser melting (SLM), for the manufacturing of various ODS alloys has been demonstrated using mechanically alloyed powder material as a feedstock [19][20][21][22][23][24]. An approach utilizing laser-processed powder material containing the metallic and ceramic components of the alloy as a feedstock also has been demonstrated. [25][26][27][28] For the LAM process of directed energy deposition (DED), also referred to as laser metal deposition (LMD), general feasibility has been demonstrated using mechanically alloyed powder material [29][30][31]. However, pronounced agglomeration and clustering of oxide nanoparticles in DED-manufactured material are observed [32,33], resulting in the dispersion of agglomerated oxide dispersoids with reduced number densities compared to L-PBF processed material. [27] The agglomeration tendencies are attributed to reduced solidification and cooling rates, which stem from larger beam diameters and lower process velocities used in DED processes compared to L-PBF processes [27] or the utility of additional preheating [29,34].
2 Experimental procedure

Powder material
The used metallic powder materials Fe20Cr4.5Al0.5Ti (wt.%) were received in an Ar-atomized state (Nanoval GmbH, Berlin, Germany) in a particle size distribution of 15-45 µm with a d 50 value of 32.1 µm (Fig. 1a) according to the manufacturer. Y 2 O 3 powder (abcr GmbH, Karlsruhe, Germany) was received in a particle fraction of 20-40 nm. Due to massive agglomeration by electrostatic forces, nanometer-sized powder material cannot be conveyed via conventional powder feeding systems, which makes the manufacturing of a powder compound consisting of micrometersized metallic powder and nanometer-sized powder necessary. Manufacturing of a powder compound was performed via mechanical alloying in the planetary mill Pulverisette 4 classic line (Fritsch GmbH, Idar-Oberstein, Germany) using milling containers and grinding balls manufactured from yttrium stabilized zirconia (YSZ). The result was a powder compound consisting of metallic particles covered with nanometer-scaled Y 2 O 3 particles on the surface. Only minor deformation of the metallic particles by the milling process was observed; hence, the powder compound was considered suitable for the used powder feed system (Fig. 1b).
EDS measurements on the surface of the powder particles shown in Table 1 confirm the presence of yttrium and oxygen on the surface of the mechanically alloyed powder material.

High-speed laser cladding (HSLC)
The HSLC process was performed using a 4-axis gantry system equipped with a fiber-coupled diode laser system LDF 2000-30 (Laserline GmbH, Mülheim-Kärlich, Germany) emitting laser radiation with wavelengths of 1025 nm and 1064 nm simultaneously. The laser beam was guided by an optical fiber and shaped via collimation (f c = 200 mm) and focusing lens (f f = 182 mm) to generate a focal spot with a diameter of approx. 0.6 mm and a top-hat beam profile. The powder material was fed by conventional disk-based powder feeding units PF 2/2 (GTV Verschleißschutz GmbH, Luckenbach, Germany) via Argon carrier gas. Before entering the powder nozzle, the powder gas stream entered a diffusor unit dividing the powder gas stream into three independent streams to homogenously feed the powder material into the continuous coaxial powder nozzle D40 (Fraunhofer ILT, Aachen), injecting the powder material into the melt pool formed by the laser beam on the substrate surface. A local argon shielding gas atmosphere was created by the coaxial powder nozzle to protect the melt pool from atmospheric oxidation (Fig. 2).
As a substrate material, steel tubes (AISI 304) with a wall thickness of 5 mm and an outer diameter of 50 mm were used, which were mounted on a lathe machine Twister FU-180 (Neureiter Maschinen GmbH, Kuchl bei Salzburg, Austria). The substrates were sand-blasted and cleaned with ethanol before mounting to avoid surface contaminations. The process parameters for the HSLC process used are shown in Table 2.

Melt pool simulations
The determination of the temperature distribution and the solidification kinetics during the HSLC process was performed via simulations. The simulation is based on a mathematical model operating with the integration of the time-dependent heat equations in conjuction with calculations of the track geometry using nonlinear Young-Laplace equations and solved with the finite element method (FEM) [27]. Temperature-dependent thermophysical data for heat capacity, thermal conductivity, and thermal expansion were taken from Kanthal APMT FeCrAl-based steel [41]. A density of 7.25 g/cm 3 [42] is

Characterization methods
The microstructural characterization was performed by standard metallographic analysis. Specimens were cut parallel to the traverse direction and embedded in a conductive resin followed by grinding with SiC paper and final polishing procedure using 1 µm diamond suspension.  Figure 3 provides LOM images of cross sections consisting of two subsequently manufactured layers for both material types. The non-reinforced coating, manufactured with Fe20Cr4.5Al0.5Ti (wt.%) material, (Fig. 3a) features full metallurgical bonding to the substrate material without any sign of delamination or defects in the fusion zone. The full double-layered coatings are found to be almost defect-free, exhibiting no cracks and low porosity, resulting in a relative density of 99.94%. The curvature of the fusion zone proves that substantial melting of substrate material occurred, resulting in an effective layer height of approx. 100 µm. The fact that the upper layer exhibits a larger layer height indicates significant remelting of the previously applied layer.

Microstructural characterization
A strong texture in building direction (BD) is observed, indicating that the main heat diffusion direction is perpendicular to the building direction. The inclination of the grain growth direction of grains starting at the substrate-cladding interface is caused by the curvature of the melt pool [43,44]. The curvature of the melt pool is directly connected with the melt pool geometry and describes the solid-liquid interface from the bottom to the surface of the melt pool during the deposition process. [45] The grain growth direction is not altered by the layer boundary (indicated by the red line), and epitaxial grain growth (indicated by purple arrows) is observed regularly. However, the growth direction changes significantly parallel to the traverse direction in the top area of the cladding (indicated by blue arrows). The abrupt change of the grain growth orientation can be explained by the curvature of the melt pool boundary and the energetically preferred growth direction, which is the smallest angle to the temperature gradient [44]. Because all these features can also be observed in the cladding manufactured with Fe20Cr4.5Al0.5Ti + 0.5% Y 2 O 3 (wt.%) (Fig. 3b), it can be concluded that no apparent change between reinforced and non-reinforced cladding could be observed on a scale that can be resolved by LOM.
The microstructure of the non-reinforced material (Fe20Cr4.5Al0.5Ti) was examined using SEM images shown in Fig. 4a and is characterized by grain boundaries (marked with yellow arrows) in the build direction and a predominantly dendritic solidification structure. Within the microstructure, no precipitations or dispersoids can be detected (Fig. 4b). The sporadically observed flakes (green circles) could be identified as remnants of the etching process and were therefore discarded in the microstructure analysis.
SEM images of the reinforced material reveal the presence of sub-micrometer scaled oxide particles homogeneously distributed inside the grains, appearing as bright white spots throughout the microstructure (Fig. 4c). The microstructure evolution is not affected by the presence of the nanometer-scaled oxide particles, which are not acting as a site for heterogeneous nucleation of the ferritic matrix. This may be attributed to the low wettability of Y 2 O 3 by iron-chromium melts [46]. The size of the nanoparticles (marked with red arrows) ranges between 50 and 150 nm (Fig. 4d). However, smaller nanoparticles may also be present, which cannot be resolved with the used SEM system. The nanoparticles are found to be significantly smaller than in PM2000 material processed via conventional DED (100-200 nm) [25] but larger than nanoparticles found in L-PBF processed with approx. 60 nm [21] and conventionally manufactured PM2000 with approx. 60 nm [47].
This process-related refinement of dispersed oxide particles can be further examined by modeling of the respective temperature distribution in the melt pool, the cooling rate, and the solidification rate encountered in the highspeed deposition process with the process parameters used ( Table 2).
Comparing the calculated process-related values for HSLC with DED and L-PBF processes processing the same alloy, indicate that the considered values of maximum melt pool temperatures, solidification rates and cooling rates in HSLC are shifted from DED toward L-PBF (Table 3). The maximum melt pool temperature directly influences the melt viscosity and may enhance nano-particle mobility in the melt pool. Therefore, low melt pool temperatures are considered favorable for preventing nano-particle agglomeration. In Fig. 5c, high-temperature gradients in HSLCformed melt pools are observed, which are believed to cause severe melt pool dynamics induced by maragoni convection, effectively preventing particle agglomeration. We hypothesize that even disruption of occasionally formed agglomerates by maragoni-induced shear forces might occur. Another factor for successful dispersion of nanoparticles is high cooling and solidification rates limiting the existence time of the melt pool significantly inhibiting diffusion controlled agglomeration mechanisms [48]. Nanoparticles dispersed in the melt pool are instantaneously captured by the advancing solidification front and integrated in the mushy zone inhibiting the nanoparticles' mobility and therefore preventing agglomeration tendencies. Low solidification rates allow nanoparticles to agglomerate in the melt pool impeding capture by the mushy zone and are consequently being pushed in front of the solidification front [49]. Additionally, advanced particle agglomeration enforces buoyancy-induced flotation effects of nanoparticles leading to slag formation on the clads' surface and therefore reduced dispersion strengthening. [50] However, a direct comparison between the powder-blown deposition techniques of DED and HSLC, and the powder bed manufacturing process of L-PBF is difficult because of the strongly differing beam diameters (DED and HSLC: 0.66 mm [25], L-PBF: 0.08 mm [27]) leading to formation of increased melt pool dimensions in DED and HSLC, altering the solidification conditions significantly. The direct comparison of DED and HSLC with identical beam diameters used proves the ability to increase cooling and solidification rates leading to improved dispersion of nanoparticles in the stainless steel matrix reflected by smaller oxide particles in HSLC (50-150 nm) compared to DED (100-200 nm) [25].  Cooling rate in the solidification zone. Negative cooling rates can be considered as heating rates by the laser irradiation. c Solidification rate plotted in transverse (TD) and building direction (BD) The chemical composition of the dispersoids in the ODS specimen could be evaluated using EDS analysis on rarely observable agglomerates, revealing enrichments of yttrium, oxygen and titanium. This indicates the formation of Y 2 Ti 2 O 7 or Y 2 TiO 5 compounds typically found in titanium-alloyed Fe-Cr-based ODS steels [51]. The formation of Y-Al-O particles, found in Al-alloyed Fe-Cr-based ODS steels [52,53], could not be observed. These findings are in agreement with DED-manufactured material. [25] However, a detailed study using high-resolution electron microscopy is necessary to confirm these results.

Mechanical characterization
The hardness profiles of both studied materials show a higher hardness of the reinforced material with hardnesses locally exceeding 300 HV0.1 (Fig. 6) proving the impact of nanoparticles on the hardness of the material. Please note that the nominal nano-particle mass loadings are different between the conventionally DED-manufactured sample (0.3 wt.%) [25,54] and the HSLC (0.5 wt.%) processed material from this study.
However, the HSLC processed non-reinforced stainless steel (blue) exhibits a hardness of approx. 250 HV0.1, which is higher compared to its DED-manufactured specimen (approx. 230 HV0.1). Grain refinement enabled by higher solidification kinetics in the HSLC processed specimen may be a reason for this behavior. It can be concluded that nanoscaled dispersed oxides increase the hardness presumably by the Orowan mechanism leading to pronounced dispersoid hardening.

Conclusion
The feasibility of using the advanced DED process of highspeed laser cladding (HSLC) for the production of the Febased oxide-dispersion-strengthened alloy PM2000 has been demonstrated. Following major findings can be drawn from this study: 1) Manufacturing of defect-free specimens with low porosity using the high-speed laser cladding (HSLC) process was successful. 2) After HSLC processing, the reinforced PM2000 powder, a homogeneous distribution of sub-micrometer scaled Y-Ti-O-structured oxides within the metallic Fe20Cr4.5Al0.5Ti-matrix (wt.%), was achieved. 3) A considerable increase of hardness (approx. 320 HV0.1) of the reinforced PM2000 material was determined exceeding the hardness of the non-reinforced counterparts (approx. 260 HV0.1).
In-depth microstructural characterization using transmission electron microscopy (TEM) to clarify the crystallographic structure, size distribution and chemical composition of the nano-scaled oxides and electron backscatter diffraction (EBSD) examinations will be part of future studies to further characterize the local grain structure. Furthermore, mechanical characterizations for the determination of tensile and high-temperature properties will be performed.

Conflict of interest
The authors confirm that there are no known conflicts of interest associated with this publication.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Fig. 6 Hardness profile over the building direction of Fe20Cr4.5Ti0.5 (blue) and Fe20Cr4.5Ti0.5 + 0.5% Y 2 O 3 (red). Error bars indicate the respective standard deviation. Hardness measurements for DED-manufactured material (light and dark green lines) are taken from [54]