Meta-structure of amorphous-inspired 65.1Co28.2Cr5.3Mo lattices augmented by artificial intelligence

Abstract

A hatching-distance-controlled lattice of 65.1Co28.2Cr5.3Mo is additively manufactured via laser powder bed fusion with a couple of periodic and aperiodic arrangements of nodes and struts. Thus, the proposed lattice has an amorphous-inspired structure in the short- and long-range orders. From the structural perspective, an artificial intelligence algorithm is used to effectively align lattices with various hatching distances. Then, the metastable lattice combination exhibits an unexpectedly high specific compression strength that is only slightly below that of a solid structure. From the microstructural perspective, the nodes in the newly designed lattice, where the thermal energy from laser irradiation is mainly concentrated, exhibit an equiaxial microstructure. By contrast, the struts exhibit a columnar microstructure, thereby allowing the thermal energy to pass through the narrow ligaments. The heterogeneous phase differences between the nodal and strut areas explain the strength-deteriorating mechanism, owing to the undesirable multi-phase development in the as-built state. However, solid-solution heat treatment to form a homogeneous phase bestows even higher specific compression strength. Furthermore, electrochemical leaching leads to the formation of nanovesicles on the surface of the microporous lattice system, thereby leading to a large surface area. A more advanced valve cage for use in a power plant is designed by using artificial intelligence both to (i) effectively preserve its mechanical stiffness and (ii) actively dissipate the generated stress through the large surface area provided by the nanovesicles.

Supplementary information: The online version contains supplementary material available at 10.1007/s42114-024-01039-6.

Keywords: Amorphous-inspired structure; Artificial intelligence; Hatching-distance-controlled lattice; Heterogeneous phase differences; Metastable lattice combination.

Fig. 1

a A photograph of the 65.1Co28.2Cr5.3Mo structure with a hatching distance of 800 µm after removing the excess powder by electrochemical leaching. b The µ-CT image of the lattice shows the pore-size distribution. c The specific surface area of the lattice measured using the µ-CT images. The core (d) was still slightly infilled with the powder due to less electrochemical leaching in this area, whereas the edge (e) was fully etched, which reveals the morphology of the nodes and arrangement of the struts. r, s The specific surface area of the lattice measured using the µ-CT images and BET analysis. t A comparison of the roughness testing results for the lattice structures with hatching distances of 80 (solid), 200, 400, 600, or 800 µm. u A comparison of the heat transfer coefficients of the lattices on the side plane. SEM images of the cellular structure at various magnifications: f–k the top plane before and after leaching, respectively, and l–q the side plane before and after leaching, respectively

Fig. 2

a A schematic illustration of the heat-transfer mechanism in the as-built 65.1Co28.2Cr5.3Mo lattice additively manufactured with a hatching distance of 800 µm before HT. b–e The top-plane images show the formation of (i) equiaxial dendrites in the nodal area, where the thermal flow was hindered by loosely-attached regions or empty spaces, and (ii) columnar dendrites in the strut area, where the thermal energy flowed easily along the entangled longitudinal ligaments during AM. (f–i) The side-plane images show fewer acicular dendrites compared to the microstructures in the solid. j–m The top-plane images show that the melt pools generated by laser irradiation disappeared and were replaced by blunt-shaped columnar dendrites within distinguishable grains from the inner core to the external surface along the direction of heat extraction, irrespective of the nodal and strut areas. n–q The side-plane images show blunt-shaped columnar dendrites comprising only the primary γ-fcc phase

Fig. 3

a, i The IQ maps; b, j the grain size distributions; c, k the grain boundary maps (red: < 5° boundary; green: > 5° and < 15° boundaries; blue: > 15° boundary); and d, l the phase maps of the additively manufactured 65.1Co28.2Cr5.3Mo lattice structure before and after HT. e, m The KAM maps (color contrast images from blue to red according to the misorientation density); f, n the IPF maps; and g, h, and o the pole figures (g and o for the γ-fcc phase and h for the ε-hcp phase) of the hatching-distance-controlled lattice structure before and after HT

Fig. 4

a–c The TEM images of the as-built 65.1Co28.2Cr5.3Mo lattice (800 µm hatching distance) show the segregates and precipitates along the interfacial boundaries between the γ-fcc and ε-hcp phases, which significantly affected the compression strength. d The low-magnification; e–f, i–j, and m–n intermediate-magnification; and g, k, and o high-resolution TEM images with the corresponding lattice fringe measurements. h, l, and p The SAED patterns of (h) the segregates and precipitates, (l) primary γ-fcc phase, and (p) secondary ε-hcp phase. The XRD patterns of the 65.1Co28.2Cr5.3Mo structures with hatching distances of 80 (solid), 200, 400, 600, or 800 µm: q before HT, showing that the small amount of the secondary ε-hcp phase in the primary γ-fcc phase increased distinctively owing to greater heat condensation in the peripheral area of the disconnected melt pools (where the thermal energy could not flow easily through the powder with low heat conductivity) than in the core area; r after HT, indicating that the lattices with the heterogeneous phases became homogeneous, thereby suppressing the formation of the secondary phases (ε-hcp, segregates, and precipitates)

Fig. 5

a, b The compression strengths of the as-built lattices gradually decreased with increasing hatching distances because of their lower relative densities and hollower structures. However, the lattices showed unexpectedly high compression strengths owing to (i) their amorphous-like structures and (ii) the presence of highly overlapped nodes with densely aligned struts. c After the removal of the powder from the lattices via electrochemical leaching, the collapse was induced along the planes when the external load was applied. This may be due to the reduction in the deformation behavior of the bound powder, as indicated by considerable smoothing of the undulations in the compression testing graphs. d Meanwhile, the lattices had higher compression strengths after HT owing to (i) the uniform formation of only the primary γ-fcc phase, irrespective of the nodal and strut areas, and (ii) the removal of the residual and thermal stresses

Fig. 6

Compression strength evaluations of the multi-modal lattice structures comprising a 2, b 4, or c 8 lattices. d AI was used to suggest the 12-lattice combination of 700/300/400/500/300/600/300 (7 in the upper block) and 300/800/300/600/300 (5 in the lower block), to achieve a yield strength of 300.0 MPa. FEM was used to calculate the yield strength of the meta-structure. Finally, the actual yield strength of the additively manufactured lattice combination was measured experimentally via compression testing

Fig. 7

Step 1: Photographs and schematics of the original part before topological optimization. Step 2: The equivalent stress gradient distribution in the original part under the applied pressure (3 MPa) and force (5000 N) to verify the solid deposition and space removal during AM. Step 3: Topological optimization to maximize the stiffness and minimize the weight of the part so that it can bear the maximum equivalent stress of 410.0 MPa (close to one-third of the yield strength (416.2 MPa) for the aged 65.1Co28.2Cr5.3Mo solid additively manufactured along the vertical direction) while reducing its weight by more than 43%. Step 4: The residual and thermal stresses caused during the recycling of hot and cold fluids become concentrated and accumulated on the main components of the valve cage. Thus, to effectively dissipate the residual heat out to the valve cage, maximize the heat exchange effect, and lower the heat accumulation, the previously designed lattice structures with various hatching distances were attached to the outer plate of the topologically optimized part. Step 5: Three lattice combinations with various hatching distances (5 lattices: 80 (solid), 200, 400, 600, and 800 µm; 8 lattices: 80 (solid), 200, 300, 400, 500, 600, 700, and 800 µm; and 15 lattices: 80 (solid), 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, and 800 µm) were arranged on three separate compartments (3, 5, and 7) with an identical number of elements (18,412) on the outer plate of the topologically optimized valve cage based on attaining a high compression strength and large surface area in the combined ratio of 50%: 50% of their maxima by using deep learning